ABSTRACT RELATIVE CONSTRUCTIONS IN MARATHI By Sharad M. Gupte The subject of this study is relative constructions in Marathi, a modern Indo-Aryan language spoken in the state of Maharashtra, India. The study begins with a review of previous analyses of restrictive and non-restrictive relative clauses in English. within the framework of transformational grammar. The Conjunction analysis is assumed for both restrictive and non-restrictive relative clauses in Marathi. Multiple headed relative clauses in this language seem to provide some internal evidence in support of the Conjunction analysis for restrictive relative clauses. It is proposed in this study that the presence of a correlative pronoun in a relative clause in Marathi serves as the surface structure clue that enables the hearer to relate the head NP to the preceding relative clause. Further. in contrast to English, in some cases Marathi makes a formal distinction between transparent and Opaque contexts. In Marathi, restrictive and non-restrictive relative clauses are not always formally distinct. Therefore. in the absence of any formal distinction between restrictive and non-restrictive relative clauses, their semantic interpreta- tion is determined by the nature and extent of the hearer's knowledge about the referent of the head NP. Some, if not all. of the non-restrictive relative clauses relate not only to the topic but also to the comment in the main clause. In such cases, the corresponding conjoined sentences fail to express the adverbial relations, e.g. cause, or concession. It is proposed in this study that in Marathi, relative clauses and relative participle phrases are derived from the same underlying structure. Further, the formation of relative participle phrases in Marathi is subject to a certain con- straint. which ensures that functional information is pre- served in the surface structure of a sentence. In addition. it is shown that relative participle phrases in Marathi play a significant role in reducing the perceptual complexity of such structural configurations as self-embedding. In the formation of relative clauses and questions. Ross' (1967) Complex NP Constraint, the Coordinate Structure Constraint. the Sentential Subject Constraint and the Pied Piping Convention are not applicable in Marathi. The reason for this is that in contrast to English, in Marathi RELATIVE CLAUSE FORMATION and QUESTION FORMATION are not reordering transformations. Lastly. in contrast to English. in Marathi the focus and relative clause constructions are formally distinct. Nevertheless, they are similar to each other in one important respect: just as a focus construction can have more than one noun phrase in focus, so also a relative clause can have more than one relativized NP i.e. multiple headed relative clauses. This cannot be just an accidental similarity: rather Schachter's (1973) notion of foregrounding neatly captures the semantic property shared by these two construc- tions. RELATIVE CONSTRUCTIONS IN MARATHI By \ 50‘ (I \ Sharad M? Gupte A DISSERTATION submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Linguistics and Oriental and African Languages 1975 ACKNOWLEDGMENTS I wish to express my sincere appreciation to the following individuals. who served on my committee: Dr. Julia Falk (chairman and thesis supervisor) Dr. Ruth Brend (member) Dr. Rachel Costa (member) Dr. John Eulenberg (consultant) Dr. Yamuna Kachru (member) Dr. Marc Rosenberg (consultant) Dr. Seok Song (member) Without their invaluable help and varied talents this study could not have been completed. Dr. Julia Falk, who also served as my academic advisor during most of my graduate study at Michigan State University, deserves special mention for her instinctive ability to offer constructive criticism or extend needed encouragement at the prOper times and in the right amounts. In addition. she has been very helpful to me in the editing and proof-reading of the manuscript. for which I am profoundly grateful. I would also like to thank Dr. Vasant Khokle, who served as a member and a temporary chairman of the committee ii prior the I enco stud (I v 9? and w I u my 3a: iii prior to his departure to India. I am also indebted to Prof. James Wang, Chairman of the Linguistics Department, for his personal support and encouragement freely given during all phases of my graduate study in this department. Finally, I would like to express a deep debt of gratitude to Sharmila, my wife, for her patience and love. and for her moral support. Special thanks are due to her for typing the earlier and the final drafts of the manuscript. I would also like to express my warm appreciation to Naren, my son, for his continued love and understanding even when Daddy was ”too busy.” Chapter by. Charm 5% ”\+ “‘3 v9 9, Chapter Chapter Chapter Chapter Chapter TABLE OF CONTENTS Introduction . . . . . . . . . . . . . . A Brief Survey of Previous Analyses of Relative Clauses in English . . . . . 2.1 The definition of relative clauses in English . . . . . . . . . . . . . 2.2 Differences between restrictive and non-restrictive clauses in English . 2.3 The underlying structure of non-restrictive relative clauses in English . ... . . . . . . . . . . 2.4 Various proposals regarding th underlying structure of restrictive relative clauses in English . . . . Relative Clauses in Marathi . . . . . . . 3.1 Some characteristics of Marathi . . 3.2 A restrictive relative clause in Marathi . . . . . . . . . . . . . 3.3 Formation of a restrictive relative clause in Marathi . . . . . 3.“ A non-restrictive relative clause in Marathi . . . . . . . . . . . . . 3.5 Survey of two previous studies on relative clauses in Marathi . . . Relative Participle Phrases in Marathi . 4.1 Formation of relative participle phrases in Marathi . . . . . . . . . h.2 Relative participle phrases and perceptual complexity . . . . . . . Relative Clause Formation and Question Formation in Marathi and Ross' Constraint 5.1 RELATIVE CLAUSE FORMATION and QUESTION FORMATION in Marathi . . . 5.2 Relative clauses in Marathi and Ross' (1967) constraints . . . . . . iv 12 39 39 so 59 72 83 9h 94 116 129 129 135 3h3pt! Bibli: V 5.3 Focus and relative clause constructions in Marathi . . . . . . . 152 5.4 Multiple headed relative clauses in Marathi . . . . . . . . . . . . . . 158 Chapter 6 Conclusion . . . . . . . . . . . . . . . . 171 Bibliography 0 O O O O I C O O O C O O O C O 0 O O O O 1 7 G decad that of Lia CHAPTER 1 INTRODUCTION The purpose of this study is to gain some understand- ing of the relativization process in Marathi. a modern Indo- Aryan language spoken in the state of Maharashtra, India. A survey of linguistic studies during the past two decades in the United States as well as in India reveals that Marathi is one of the least studied modern Indo-Aryan languages. For example, in the United States only three doctoral dissertations have been written since 1958: Kelkar (1958), Apte (1962) and Khokle (1969). Kelkar (1958) deals with the phonology and morphology of Marathi within a structuralist framework. Apte (1962) is a linguistic description of Marathi within the framework of Syntactic Structures ( Chomsky, 1957). The author gives phrase structure rules. a series of optional and obligatory transformations. morphophonemic rules. morpheme structure rules and phonological rules to generate 'kernel' and 'derived' sentences. This study is not in any way an in-depth analysis of any aspect of Marathi syntax. Khokle (1969) applies the two models of distinctive 1 featun to abm harath phenol clause which harath Speake Withir 2 features (i.e., the binary features and the n-ary features) to about 115 redundancy rules and phonological rules of Marathi and shows that the binary approach yields a simpler phonological description. Recently there appeared two studies on relative clauses in Marathi (i.e., Junghare (1973) and Kelkar (1973)), which are discussed in section 5 of chapter 3 of this study. The present study, based on the standard dialect of Marathi, known as the 'Puneri Boli' and spoken by educated speakers from Poona (Maharashtra state, India), is done within the framework of transformational grammar. Chapter 1 presents the organization of this thesis. Chapter 2 summarizes previous analyses of restrictive and non-restrictive relative clauses in English. Chapter 3 ennumerates some typological characteristics of Marathi. Furthermore, this chapter discusses the formation of restrictive and non-restrictive relative clauses in Marathi. Also, in the same chapter there is a discussion about two previous studies on relative clauses in Marathi. Chapter 4 describes the formation of relative participle phrases in Marathi and the role these phrases play in reducing the perceptual complexity of such structural configurations as self-embedding. Chapter 5 shows that Ross' (1967) Complex NP Constraint, Sentential Subject Constraint, Coordinate Structure Constraint and Pied Piping Convention are not 3 applicable in Marathi. Furthermore, in this chapter it is shown that the relative clause construction and the cleft construction in Marathi do not lend support to the Promotion analysis (Schachter, 1973). Lastly, in the same chapter there is a discussion about focus and multiple headed relative clause constructions in Marathi. It is argued that the semantic difference between single headed and multiple headed relative clauses in Marathi can be explained in terms of the notion of foregrounding (Schachter, 1973). Chapter 6 summarizes the main conclusions of this study. CHAPTER 2 A BRIEF SURVEY OF PREVIOUS ANALYSES OF RELATIVE CLAUSES IN ENGLISH 2.0. This chapter is divided into the following sections: 2.1. 2.2. 2.3. 2.“. The definition of relative clauses in English Differences between restrictive and non-restrictive relative clauses in English The underlying structure of non-restrictive relative clauses in English Various prOposals regarding the underlying structure of restrictive relative clauses in English 2.1. The definition of relative clauses in English In recent years the study of relative clauses in English within the framework of transformational grammar has been focused on such issues as the differences between restrictive and non-restrictive relative clauses, their underlying structures, definite and indefinite noun phrases containing relative clauses, quantified noun phrases containing relative clauses, and constraints on movement of elements from within relative clauses. Stockwell et al.(1973, p.421) define a relative L; 5 clause in English as follows: A sentence embedded (in surface structure) as modifier of an NP, the embedded sentence having within it a NH: . pronominal replacement for a deep structure NP which IS in some sense identical with the head NP, is a relative clause. LThus, for example, the underlined part in each of sentences (2.1.1) and (2.1.2) is a relative clause. (2.1.1) The man who came to dinner left early. (2.1.2) My brother, who came to dinner, left early. Schachter (1973, p. 43) says the following about the relativization process: This fact points to whatifisprobably the very raison d'Stre of relativization: to provide names for, or ways of designating thernultitude of entities that people wish to talk about, but for which there is no established single- noun designation (or, perhaps, where the people do not know, or do not choose to use, such a designation). In other words, a relative clause is a syntactic device which is used to express certain properties of a noun for which there may not be a single lexical item. For example, consider the following sentences: (2.1.3) a. The man who was old died. b. The old man died. The property of the NP 'the man' in sentence (2.1.3) a which is expressed in the relative clause 'who was old' may be expressed by using the single lexical item 'old' as shown in sentence (2.1.3) b. However, notice that there are no one-word equivalents of the relative clauses in the following sentences: 6 (2.1.4) The girl who danced on the table is sick. (2.1.5) The dog that the child waspetting died. There are two types of relative clauses in English: restrictive and appositive (or non-restrictive). These two types of clauses differ functionally in that they stand in different relations to the NP they modify. To understand this consider the following "semantically based--- and syntax independent--- notion" of a restrictive relative clause (RRC), as given by Keenan (1972, p. 169): In the simple case a RRC is a way of specifying an individual (or set of individuals) in two steps: first, a larger set of individuals is specified--- hereafter referred to as the domain of the RRC, and then that set is restricted to those members which have the property expressed by a certain sentence--- hereafter, the restricting sentence. In the present study I assume the above definition for the analysis of restrictive relative clauses in Marathi. Consider sentences (2.1.1) and (2.1.2) again, reproduced below: (2.1.1) The man who came to dinner left early. (2.1.2) My brother, who came to dinner, left early. The underlined clause in (2.1.1) Specifies an individual 'the man' out of a larger set of individuals 'men', thus restricting the scope of the modified NP: hence it may be called a restrictive relative clause. On the other hand, the underlined clause in (2.1.2) does not restrict the scope of the modified NP 'my brother' since it is already specific: hence this may be called a non-restrictive relative clause. 2.2 Differences between restrictive and non-restrictive relative clauses in English1 There are several differences between restrictive and non-restrictive relative clauses in English: (1) Non-restrictives are separated by intonation commas: whereas restrictives do not require them. (ii) Non-restrictives may modify proper nouns. but restrictives may not do so unless determiners precede the proper nouns, since the intrinsic function of proper nouns - to identify a unique individual without mentioning properties that would single out this individual from within a larger set - is incompatible with the above described function of restrictive relative clauses. (2.2.1) Bill, who came to dinner, left early. (2.2.2) *Bill who came to dinner left early. (2.2.3) The Bill who came to dinner left early. (iii) Restrictives allow that as a relative pronoun, but non-restrictives do not: (2.2.4) The man that came to dinner left early. (2.2.5) *John, that came to dinner, left early. (iv) Restrictives may modify NPs having any or 29: 8 non-restrictives do not. (2.2.6) Any mango which is raw is sour. (2.2.7) *Any mango, which is raw, is sour. (2.2.8) No mango which is raw is sweet. (2.2.9) *No mango, which is raw, is sweet. (v) Non-restrictives may contain performatives; restrictives do not. (2.2.10) The book, which I hereby promise to send you, is banned in Europe. 2.3 The underlying structure of non-restrictive relative clauses in English There is a fair amount of discussion about the under- lying source of both non-restrictives and restrictives. Lees (1963) and Smith (1964) pr0pose that both restrictive and non-restrictive relative clauses are derived from underlying sentential embedding configurations. On the other hand, the similarities between non-restrictives and conjunc- tions2 have led the following linguists to propose that non-restrictive relative clauses in English are derived from underlying conjunctions: Aissen (1972), Annear (1967, 1970b). Bach (1968), Drubig (1968), Jacobs and Rosenbaum (1968), Lakoff (1966, 1968), Postal (1967), Ross (1967), Schachter (1973)- For example, Lakoff (1968) argues as follows: 3A Nu~ 9 (i) There is a paraphrase relation between non-restric- tives and corresponding conjunctions, e.g. (2.3.1) a. Drug manufacturers, who are rich, are thieves. b. Drug manufacturers are thieves and they are rich. (ii) The conjunction analysis explains why noun phrases having Q3 or any do not take non—restrictives: Notice that sentences (2.2.7) and (2.2.9) are ungrammatical. The conjunction analysis predicts that the correSpond- ing conjunctions derived from the common source will be ungrammatical, and in fact it is the case: sentences (2.3.2) and (2.3.3), which correspond to sentences (2.2.7) and (2.2.9), respectively, are ungrammatical: (2.3.2) *Any mango is sour and it is raw. (2.3.3) *No mango is sweet and it is raw. (iii) The following argument in support of the conjunction analysis for non-restrictives involves the scope of negation. Consider the following sentence: (2.3.4) John denied that Harry, who robbed the bank, shot Bill. Following an embedding analysis for non-restrictives (e.g. Lees (1963), Smith (1964)), the (approximate) deep structure of (2.3.4) would be as given in (2.3.5). 10 (2-3-5) So NP VP John /\ Y /NP\ denied it Si /\ NP EE 4”,/””’\\\‘\\\\‘ shot B111 MP S2 A Harryi Harryi robbed the bank Note that in sentence (2.3.4) what is denied is Hg;ry_shot Bill and not Harry robbed the bank. However, the deep structure in (2.3.5) shows that both Harry shot Bill and Harry robbed the bank are within the scope of the verb ggny. The conjunction analysis, on the other hand, yields the correct reading: (2.3.6) S and S1 . isz {7 fi=——- Harryi robbed the bank NP P JoHn \\\\\\\\\ Y NP denied I’,/””’~\\\“~\\“ it S3 Ross (1967, p. 240) gives the following counterexample to the claim that non-restrictives are derived from second () (U .117 L) U) ’1 11 conjuncts of the underlying conjoined structure: (2.3.7) Is even Clarence, who is wearing mauve socks, a swinger? Note that a conjunction paraphrase of (2.3.7) does not exist: (2.3.8) *Is even Clarence a swinger, and he is wearing mauve socks? Imperatives behave similarly in this respect: (2.3.9) Call my brother, who is downstairs. The conjunction paraphrase of (2.3.9) does not exist: (2.3.10) *Call my brother and he is downstairs. Ross, therefore, proposes the following modification: Non-restrictives arise not from conjoined sentences but from the corresponding sequence of two independent sentences.3 Thus, (2.3.11) and (2.3.12) underlie (2.3.7) and (2.3.9), respectively: (2.3.11) Is even Clarence a swinger? Clarence is wearing mauve socks. (2.3.12) Call my brother. He is downstairs. However, even Ross' revised proposal, namely that non-restrictives arise from a sequence of two independent 12 clauses, is not totally satisfying in certain cases: (2.3.13) a. John, who was born rich, died as a pauper. b. John died as a pauper. He was born rich. (2.3.14) a. Nobody helped John, who always helped others. b. Nobody helped John. He always helped others. Note that the (a) and (b) sentences are not exact paraphrases of each other: the sense of irony or paradox, which is present in the (a) sentences, is not conveyed by the (b) sentences. So it is obvious that the deep structures of the (a) sentences in (2.3.13) and (2.3.14) contain more than just a sequence of two independent clauses. The speaker of these and such other sentences is not merely stating some facts but is emotionally involved in them (i.e. he may be surprised, happy or sad to say what he says). More will be said about this in chapter 3. 2.4 Various proposals regarding the underlying structure of restrictive relative clauses in English Turning to restrictive relative clauses in English, various proposals have been made in recent years regarding their underlying structures. These proposals are summarized as follows (Stockwell et al. 1973, pp. 423-441): 2.4.1 The Art-S Analysis: This analysis, which was originally proposed by Lees (1963) and Smith (1964), is followed by Chomsky (1965). Hit det 13 Within this analysis restrictives are a constituent of the determiner in the deep structure. (2014‘0101) SO / — NP VP A ,/’/////\\\\\ is a writer Det N l ’//////’/\\\\\ man Ant S1 the you met the man 'The man you met is a writer.‘ The node S1 in the above deep structure underlies a restrictive relative clause. The proponents of this analysis thus suggest that the function of a restrictive relative clause is similar to that of a determiner: to specify the scope of reference of the NP it modifies. According to Lees (1963) the definite article and the relative clause form a discontinuous constitient (The--- you met). The prOponents of this analysis give the following evidence to show that determiners and restrictives share certain syntactic properties: there are some words that occur only if there is either a relative clause or some kind of demonstrative determinerzkind, manner, place, time, way (Stockwell et al. 1973, p. 424): (2.4.1.2) *He did it in a/the way. (2.4.1.3) He did it in a certain way. it: 14 (2.4.1.4) He did it in that way. (2.4.1.5) “He did it in a/the way that I prescribed. Jackendoff (1968) shows that there are some noun phrase constructions where the definite article is impossible unless accompanied by a relative clause: (2.4.1.6) *the piece of the cake (2.4.1.7) the piece of the cake that I didn't eat. (2.4.1.8) *He greeted me with the warmth. (2.4~1.9) He greeted me with the warmth I expected of him. The Art-S analysis, however, runs into problems of stating the identity condition between coreferential NPs, especially in sentences with self-embeddings (Stockwell et al- 1973. p. 425): h: t5 .F'f 15 (2.4.1.10) SO NP1 VP A (If/,I”’\“~\\‘\\ won the race Det N1 /\ horse Afit S1 the /\ NP2 VP A ”,,,/”’~\“\\\\‘ finished fast Det N2 ’/,/’///A\\\\\\\\ horse Art 82 the /\ NP3 VP A ’///,/,"~\\\\\\\‘ started late Det N I I 3 the horse In (2.4.1.10), if the identity condition is said to held between N1 and N2 then the following ungrammatical sen- tence is generated by the rules of relative clause formation: (2.4.1.11) *The horse that that started late finished fast won the race. Notice that the identity condition cannot be said to hold between NP1 and NP2 of (2.4.1.10) since the former contains 31’ whereas NP2 does not contain that S. If the identity condition is said to hold between the ART and its head noun of the main clause and the ART and its head noun of the subordinate clause (i.e. between the How EC CC 16 ART and N1 on the one hand, and the ART and N2 on the other hand), then the problem disappears and the following sentence is generated: (2.4.1.12) The horse that started late that finished fast won the race. However, note that in the above sentence the self-embedding of (2.4.1.10) is stacked. 2.4. 2 The NP-S Analysis: This analysis has been assumed in Jacobs and Rosen- baum (1968), Lakoff (1968), Langendoen (1969) and Ross (1967). Within this analysis a sentence is embedded into a noun phrase in the deep structure of a relative clause construction. (2.“. 2.1) /§K NP1 VP A f”,,z”’\“\\\\‘\‘ is a writer NP2 Si /\ /\ Art N NFB VP the mAn you ////~\\\\ met ///“‘~\‘\‘ Art N the man ‘The man you met is a writer.‘ Notice that this analysis is free from the problems of the identity condition, referred to in the Art-S analysis: the id: hpJ an: re 11 :10 ed th 17 identity condition can be said to hold between NP2 and NP” Of (2.4.2.1). However, there are some serious problems with this analysis. For example, one of the problems has to do with relative clauses with generic NPs. (2.4.2.2) All the mangoes which are raw are sour. (2.4.2.3) All mangoes are raw. (2.4.2.4) Every mango which is raw is sour. (2.4.2.5) Every mango is raw. Note that sentences (2.4.2.3) and (2.4.2.5) are not entail- ed by sentences (2.4.2.2) and (2.4.2.4),respeqtively. Yet the condition of coreferentiality between the head NP and its coreferential NP within the embedded clause requires that the latter sentence be derived from the following underlying structures: (2.4.2.6) All mangoeé] [All mangoes are raw] NP NP S S 1 1 O [are sourP‘] VP V So (2.4.2 .7) Every mangcfl [Every mango is raw] NP NP S 81 So [is song] VP V So Thus sentences (2.4.2.2) and (2.4.2.4) show that within the O.» «xv :— \Q. 18 NP-S analysis the following constraint is needed in the case of relative clauses with generic NPs ( Stockwell et al. 1973: p0 “29): ---while the head NP can be generic, the shared NP of the relative clause cannot be, since a generic paraphrase cannot be entailed by the shared NP of any relative clause. Notice that this constraint in effect strengthens the requirement of coreferentiality. Alternatively, as proposed by Garden (1967),the NP includes neither universal quantifiers nor generic articles at the time the identity condition applies. In other words, they arise as main verbs of higher clauses in the deep structure. It has been proposed ( Bower, 1964) that generic sentences be derived from conditionals, since there is a paraphrase relation between them: (2.4.2.8) A man who earns $25,000 a year usually considers himself underprivileged. (2.4.2.9) If a man earns $25,000 a year, he usually considers himself underprivileged. (2.4.2.10) Any linguist who reads Joos will find that Chomsky is right. (2.4.2.11) If he reads Joos, any linguist will find that Chomsky is right. There is a paraphrase relation between sentence (2.4.2.8) ar dc SE SE CC sc th ti ar Pr th ma re 19 and sentence (2.4.2.9) and also between sentence (2.4.2.10) and sentence (2.4.2.11). However, there are some generic sentences (Jacken- doff,1968) for which there are no corresponding conditional sentences, e.g. (2.4.2.12) A beaver builds dams. (2.4.2.13) * If something is a beaver, it builds dams. In this connection, Stockwell et al.(1973, p. 430) say: But it is not necessary to claim that all generic sentences have conditional paraphrases, or that all conditional sentences have relative clause paraphrases. The only claim is that sentences of the form If Generic NPi NPm then Generic NPi VPn are the sole source of relative clauses of the form Generic NPi that VPm VPn Jackendoff (1968) suggests that the paraphrase rela- tionship that holds between relative clauses in generic NPs and conditional sentences tends to support the Art-S ana- lysis: in this analysis restrictive relative clauses are constituents of the determiner: the conditional clauses arise in the presentence. Under the base rule schema pr0posed by Jackendoff (1968) both the determiner and the presentence positions generalize as Spec_. And therefore the projection rule which interprets conditional clauses may be assumed to apply in the interpretation of restrictive relative clauses. However, as already mentioned, since not all generic 20 sentences have paraphrases with conditional sentences, the Art-S analysis is not by any means superior to the NP-S analysis. 2.4.3 The Promotion Analysis: This analysis, which was originally proposed by Brame (1968), is followed by Schachter (1973): (2.4.3.1) So /\ NP VP /\ m Det NOM Aft NOM S the A /1\.. NP VP vim /\ V NP mgt Det NOM the man 'The man you met is a writer.‘ When the Relative clause transformation applies, a cepy of the constituent Egg of the embedded sentence is inserted into a matrix sentence where it replaces a dummy symbol. In other words, the noun mag is promoted from the embedded S into the main S. Next, within the embedded S, the constituent that has been copied is replaced by a relative pronoun and is moved to clause initial position. 21 Schachter (1973) gives the following evidence from Brame (1968) in support of this (promotion) analysis: The first example concerns the behavior of certain idioms like make headway in relation to relative clause constructions: (2.4.3.2) We made headway. (2.4.3.3) *(The) headway was satisfactory. (2.4.3.4) The headway that we made was satisfactory. Normally, the noun headway occurs as the object of mgkg, However, in a relative clause construction (e.g. (2.4.3.4)) it may occur in other positions. This poses a problem for both the Art-S analysis and the NP-S analysis: how is one to generate sentences like (2.4.3.4) while dis- allowing sentences like (2.4.3.3)? Under the Promotion analysis the underlying structure of (2.4.3.4) will be: (2.4.3 .5) so N/\v is satIsfacIory Det /NOM\ Ait NOM S1 VP the /\ [:4 NF Aux Y PP we Past make MOM headway Another example concerns the behavior of the process of pronominalization in relation to relative clause con- structions: in non-relative clause constructions, pronomi- nalization generally takes place in a forward direction, w'ne pr< 22 whereas in certain relative clause constructions it proceeds in the opposite direction: (2.4.3.6) John thinks that Mary has an unfavorable opinion of him. (2.4.3.7) *Hei thinks that Mary has an unfavorable opinion of Johni. (2.4.3.8) The opinion of him that John thinks that Mary has is unfavorable. (2.4.3.9) *The opinion of Johni thathei thinks that Mary has is unfavorable. The underlying structure of (2.4.3.8) within the Promotion analysis will be: (2.4.3.10) 30 N/\ VP /\ A is unfavorable Det NOM Art NOM S I 1 MP VP Johnl /\ Y "NP thinks ‘/////\\\\\ it 82 A MP VP Mary /\ Y NP has ,//~\\\\ Det NOM / Art PP ' i an opinion Prep 1 of John 1 At the 81 level, the pronominalization rule applies in a forward direction, pronominalizing the second occurrence of John. Next, the nominal opinion of him is promoted into the matrix sentence. where it replaces the underlying dummy nominal. However, there are some problems with the Promotion analysis too: Since in this analysis the relative constructions are formed by promoting a nominal from an embedded sentence into a matrix sentence, it is possible to form more than 011‘ st: in 24 one relative clause construction from a single underlying structure, depending on the number of promotable nominals in the embedded sentence (Schachter 1973, p. 38): (2.4.3.11) S //0\ MP VP A N V NP ' I I admire th NOM Art NOM the [£5 ‘!/’/~P“~\“-' an earl gave a pearl to a girl (2.4.3.12) I admire the earl that gave a pearl to a girl. (2.4.3.13) I admire the pearl that an earl gave to a girl. (2.4.3.14) I admire the girl that an earl gave a pearl to. Thus, since the embedded sentence in (2.4.3.11) has three promotable NPs. three different relative clause constructions can be formed from it (i.e. (2.4.3.12) through (2.4.3.14)). Within the standard theory it should not be possible to derive three semantically different sentences from the same underlying structure. - “sf—fl -“' 2.4 11:: H N fif‘. ”,4 25 2.4.4 The Conjggction Analysis: Within this analysis it is assumed that a restrictive relative clause construction arises from an underlying sequence of two or more clauses (e.g. Annear(1967, 1970a, 1970b), Drubig (1968), McCawley (1970, 1974), Postal (1967)). Each (1968) proposes that the referential indices which occur with nouns in the Aspects model (1965) be re- placed by a system of operators and variables. Following this proposal, therefore, the underlying structure of sentence (2.4.4.1) will be as given in (2.4.4.2). (2.4.4.1) I know three boys who have beards. (2.4.4.2) There exist three 3 such that (X is boy) (I know X) (X has a beard). McCawley (1970) follows Bach in that he separates the utterance into a 'prOposition' and a set of noun phrases. McCawley (1970) seems to agree (to the extent to which one can make sense of his tree diagrams) that restrictive relative clauses should be derived from something very like a conjoin- ed structure. Consider the following sentence (McCawley 1970: P- 1714')8 (2.4.4.3) John said that he had seen the woman who lives at 219_Main gireet. The above sentence is ambiguous: in one case John is responsible for the underlined relative clause, that is he actually utters it. Whereas in the other case the speaker 26 provides the information stated in the relative clause. McCawley (1970, p. 175) proposes the following two under- lying structures for the two meanings of (2.4.4.3): (2.4.4.4) S Proposition NP:x1 A I x1 said S John Proposition PNszz x1 saw x2 the woman who lives at 219 Main Street (2.4.4.5) S Proposition NP:x1 NP:x2 x1 said S John the woman who lives at 219 Main Street Proposition X1 saw X2 Postal (1967, p. 18) says: Since appositives come from coordinate structures and appositives and restrictives share many similarities, these similarities are at least weak arguments for the coordinate character of restrictives. The similarities that Postal (1967, p. 18) refers to are as follows: (I) both end up next to another NP (II) both make use of ED marking of pronominal forms and the special fronting word order associated with this (there are differences of course, that, etc.) (III) both can be reduced to yield pronominal modifiers (IV) neither can occur next to an NP which is (+Anaphoric). a feature assigned by the rules of pronominalization to that one of two referentially identical NPs which is getting changed.... 27 However, as Postal himself admits, these similarities are in fact weak arguments. Because along with these similarities there are some differences between these two types, which are mentioned in section (2.2). Postal (1967, p. 8) gives the following examples as evidence in support of the Conjunction analysis: (2.4.4.6) Charley assumed that the book which was burned was not burned. The above sentence is syntactically ambiguous: on one read- ing, Charley assumed that a certain book had not been burned when in fact it had been. 0n the other reading, Charley assumed a contradiction. A sentence such as (2.4.4.6) poses a problem for any kind of embedding analysis. For example, following the NP-S analysis, the underlying structure of this sentence will be as given in (2.4.4.7): (2.4.4.7) S O N VP Charley ,Lv"’*‘-‘~ Y NP assumed ,La”“-~ it S1 NP—f fiVP A N§"'flflr—~PPP“7§ was not burned bo<|)k.l booki was Eurnea But notice that the underlying structure in (2.4.4.7) fails to express two meanings of (2.4.4.6). In other words, within 1“e 01‘ Cl 28 the NP—S analysis it is not possible to structurally repre- sent two meanings of sentences like (2.4.4.6). 0n the other hand, the Conjunction analysis yields the correct solution: the underlying structures in (2.4.4.8) and (2.4.4.9) represent two meanings of (2.4.4.6) (Annear, 1970t0- (2.4J+.8) s S S 1 A ’,,r””’fl\\‘\\\\\\ booki was burned NP VP Cha‘rley /\ V NP I assumed I S3 l; ‘- booki was not burned (24.4.9) so NF VP V NP assumed I //E1\ S 82 A A booki.was burned booki was not burned One of the problems with the Conjunction analysis of restrictive relative clauses is how to determine the sub- ordinate status of the relative clause with respect to the clause in which it is embedded. In this connection, Annear (1970b, p- 44) says: - If" .,- 29 The choice of the clause to become the relative clause correlates with certain presuppositions on the part of the speaker about what the hearer knows, and accordingly with the choice of the determiner. For example, consider the following sentences (Annear 1970b, p. 44): (2.4.4.10) I met a girl who speaks Basque. (2.4.4.11) A girl I met speaks Basque. (2.4.4.12) The girl I met speaks Basque. (2.4.4.13) I met the girl who Speaks Basque. According to Annear (1970b, p. 43), underlying the above sentences is a structure like (2.4.4.14): (2.414.14) (I met girl ) ( girl speaks Basque ) If the Speaker presupposes that the hearer knows neither about his meeting a girl nor about a girl's speaking Basque, sentence (2.4.4.10) or (2.4.4.11) will result. If, on the other hand, the speaker assumes that the hearer knows about his meeting a certain girl, sentence (2.4.4.12) will result, whereas if the speaker presupposes that the hearer knows about the girl who speaks Basque, sentence (2.4.4.13) will result. One of the possible objections to deriving both restrictive and non-restrictive relative clauses from conjunctions is that this assumption may force us to derive sentences of both types, which have different meanings, 30 from identical sources. According to Annear (1970b), how- ever, this objection is valid only in the case of sentences with quantifiers in them, e.g. (Annear 1970b, p. 50): (2.4J+.15) Three boys who had beards were at the party. (2.4.4.16) Three boys, who had beards, were at the party. Sentence (2.4.4.15) asserts that there were three boys all of whom both attended the party and had beards, whereas sentence (2.4.4.16) asserts that there were three boys at the party. In other cases, Annear (1970b, p. 49) argues, it seems that once again the differences between restrictive and non-restrictive relative clause sentences are not of the sort that ought to be represented structurally: instead they are differences representing a speaker's decision about how to present to the hearer information present in the underlying representation. Consider the sentences (Annear 1970b, p. 49): (2.4.4.1?) The boy, who works at the library, is majoring in philosophy. (2.4.4.18) The boy who works at the library is majoring in philosophy. In sentence (2.4.4.17) the speaker assumes that the boy is already known to the hearer. In this sentence both of the clauses add some information about that boy. 0n the other 31 hand, in sentence (2.4.4.18) the speaker assumes that the hearer knows about the boy who works at the library. In this sentence the main clause adds the information-which the speaker assumes to be new to the hearer. The discussion thus far suggests that according to Annear (1970b) in the case of a restrictive relative clause sentence at least, the embedded clause contains the pre- supposition of that sentence . In recent years some generative semanticists have proposed that the presupposition(s) of a sentence be represented in the logical structure of that sentence. For example, Morgan (1969) says: In a theory in which semantic representation is the deepest representation, presupposition will have to be represented.(p. 169) Rather, the facts show that the relationship between unuttered presuppositions and the sentence with which they are associated is exactly the same as that between a left- conjoined sentence and the conjuncts which follow it. By all indications, presuppositions are somehow conjoined to the left of the performative. (p. 174) However, Morgan (1973)rejects his own earlier analysis: For one thing, presupposing is not really performa- tive in the usual sense. By saying a sentence which has X as a presupposition, one does not thereby presuppose X. Presupposing is a state, not an act or event. Stockwell et al.(1968) argue that the correct ana- lysis of stacked relatives requires that restrictive rela- tives be derived from underlying embedded structures. In a 32 stacked relative clause construction the NP'contains more than one relative clause, which can be interpreted in such a way that each successive relative clause from right to left modifies the meaning of the head noun plus the preceding relative clauses (Annear 1970a, p. 27) .e.g. (2.4.4.19) The car that cost less than a hundred dollars that didn't break down after a hundred miles was a bargain. The stacked interpretation for (2.4.4.19) is: (2.4.4.20) Out of all the cars that cost less than a hundred dollars the one that didn't break down after a hundred miles was a bargain. Within the NP-S analysis, the underlying structure of (2.4.4.19) will be as given in (2.4.4.21): (2.4.4.21) so A NP VP A I’,,/””~P“~\\“ was a bargain NP S1 ,/’/””/’\\\\\\\\~ car. didn't break down after a hundred miles NP 82 cari cost less than a hundred dollars Det N Art car I the how int str for onl 8C in «REHEWHPSUAUC Pa .1: 33 However, Annear (1970a) rejects the proposal that stacked interpretations should be explained in terms of stacked structures. One of the reasons for this rejection is that for many speakers the stacked interpretation is possib1e< only if one of the clauses is streSsed. Otherwise they get a conjoined interpretation, e.g. The car that cost less than a hundred dollars and that didn't break down after a hupdred miles was a bargain. In other words, a stacked interpretation can be explained in terms of stress dynamics instead of a stacked structure, and therefore, it does not pose any problem for the Conjunction analysis. Ross and Perlmutter (1970) come up with the follow- ing sentence: (2.4.4.22) A man entered the room and a woman went out who were quite similar. The authors raise a question, namely what is the antecedent of the relative clause in the above sentence? They argue (1970:P- 350)‘ Neither of these singular noun phrases i.e. a man or a woman can serve as the antecedent of a relative clause whose predicate (similar) requires an underlying plural subject, and whose verb (were) is inflected to agree with a plural subject in surface structure. The only possible antecedent of the relative clause in (3) [i.e. sentence (2.4.4.22) above] would seem to be the discontin- uous noun phrase a man----—---(and) a woman. But how can a discontinuous noun phrase be the antecedent of-a relative clause? No analysis of relative clauses that has yet been proposed in the theory of generative grammar is able to account for sentences like (3). Their existence thus presents the theory with a new paradox. 34 However, Annear (1970 a, p. 57) argues that sentences such as (2.4.4.22) pose no problem for the Conjunction analysis, since within this analysis the underlying structure of this sentence will be: (2.4.4.23) (man entered room) (woman went out) (man and woman were quite similar) Notice that none of the above proposals regarding the underlying structure of restrictive relative clauses in English is free from problems. Nevertheless, it appears that a case can be made for the Conjunction analysis as follows: There is almost universal agreement among trans- formationalists that nonrestrictive relative clauses are derived from underlying conjunctions. And since non-restric- tive relative clauses share many formal similarities with restrictive relative clauses (Postal, 1967). it is only reasonable to look for a general analysis covering both types of relative clauses. As mentioned in section 4 of chapter 3, restrictive and non-restrictive relative clauses form.a continuum in regard to their function of identifying a referent of the head NP. This fact tends to support Annear's (1970b. p. 49) contention that differences between both types arise due to "a speaker's decision about how to present the hearer information present in the underlying representa- tion." She, therefore, argues that the differences between these two types of relative clauses should not be represent- ed structurally. The developmental studies dealing with the 35 ontogeny of relative clause formation (e.g. Baird (1973), Slobin and Welsh (1973), M. Smith (1974)) seem to indicate that children acquire the ability to process conjoined sen- tences prior to the acquisition of relative clause formation. M. Smith (1974, p. 654) says: A provisional conclusion is that there are parallels between Thompson's [i.e. Annear's'] conjunct analysis and developmental data. That is, at the outset, the distinction between relative clause sentences is tranSparent (or super- fluous) and, therefore, not crucial: relative clause sen- tences are analyzed by the child as being sentential con- juncts (equally prominent) with coreferential NP'S. Only later, when the presupposition-assertion distinction becomes apparent, does the superficial distinction between relative clause sentences become (or begin to become) opaque. In view of the above, following Annear (1970b),it is assumed in this study that in English as well as in Marathi, both restrictive and non-restrictive relative clauses are derived from two or more underlying clauses which are unordered with respect to one another. The Speaker's assump— tion about the nature and extent of the hearer's knowledge of a Speech act determines the subordinate status of.a clause as well as the restrictive vs. non-restrictive interpretation in a relative clause sentence. Further, in the case of rela- tive clause sentences without quantified head NPS, the restrictive and non-restrictive relative clauses (e.g. sentences (2.4.4.17) and (2.4.4.18)) are derived from iden- tical structures, while in the case of relative clause sentences with quantified head NPS, the restrictive and non-restrictive relative clauses (e.g. sentences (2.4.4.15) and (2.4a4.16)) are derived from non-identical structures. 36 Next, at some stage of their derivation, following the determination of the subordinate status of a clause, the relative clauses in English and Marathi assume language- specific structural configurations: in English the relative clause is embedded in the main clause, immediately follow- ing the head NP, whereas in Marathi the relative clause is preposed to the main clause, immediately preceding the head NP. Multiple headed relative clauses in Marathi, discussed in chapter 5, seem to provide some internal evidence in support of the Conjunction analysis for restrictive relative clauses (see section 4 of chapter 5). In summary, section 1 presents the "semantically based-- and syntax independent" definition of a restrictive relative clause which is assumed in this study for the analysis of restrictive relative clauses in Marathi. Section 2 ennumerates some syntactic differences between restrictive and non-restrictive relative clauses in English. Section 3 describes the Conjunction analysis for non-restrictive relative clauses in English. It is argued in this section that this analysis is not adequate in certain cases. Section 4 summarizes various proposals regarding the underlying structure of restrictive relative clauses. The Conjunction analysis is assumed in this study for both restrictive and non-restrictive relative clauses in Marathi. CHAPTER 2 FOOTNOTES M. Smith (1974, p. 647) says the following about the difference between restrictive and non-restrictive relative clauses in English. Perhaps the clearest formal insight into the restrictive-nonrestrictive distinction is prOVided byla system of natural logic as applied to linguistic analysis (Lakoff, 1970). The implication is that restrictive relatives function to identify a variable NP (variable referent) while nonrestrictive relatives refer to a constant NP (constant referent), hence the information supplied by a restrictive relative as Opposed to a nonrestrictive relative is crucial to disambiguation. Bach (1968) observes that there are languages which are~ lacking in non-restrictive relative clauses. The speakers of such a language use conjoined sentences to convey the message, which in another language would be conveyed by means of non-restrictive relatives. Morgan (1975) argues that in some cases at least, the content of the clauses requires a certain temporal order between them, e.g. (i) My uncle John, who, now that I've mentioned his name will no doubt appear, is a magician. (ii) This man, whose name I haven't told you yet, is John Smith, a friend of mine. 37 38 Morgan says (p. 438). "Looked at abstracted from time, (4) and (6) [i.e. sentences (i) and (ii) above] make no sense, or are self-contradictory." CHAPTER 3 RELATIVE CLAUSES IN MARATHI 3.0 This chapter is divided into the following sections: 3.1 Some characteristics of Marathi' 3.2 A restrictive relative clause in Marathi 3.3 Formation of a restrictive relative clause in Marathi 3.4 A non-restrictive relative clause in Marathi 3.5 Survey of two previous studies on relative clauses in Marathi 3.1 Some characteristics of Marathi Marathi, an Indo-Aryan language, is surrounded by two other Indo-Aryan languages, Gujarati and Hindi, and two Dravidian languages, Kannada and Telugu. Due to the geogra- phical proximity of these Dravidian languages, Marathi has assimilated several phonological and syntactic properties that are peculiar to the Dravidian language-family. The proliferation of relative participle phrase-constructions in Marathi is a case in point (for discussion see chapter 4). Marathi is a highly inflected language - - a language in which grammatical categories such as articles, demonstratives, adjectives, nouns and verbs are marked with inflectional endings. The nouns are marked for gender, number and case. There are three genders in Marathi: masculine, feminine and 39 40 neuter. The relations among constituents are expressed in Marathi by postpositions, most of which are bound formatives, e.g. (3.1.1) Root + Gender + Number Noun stem mulag- + ~52 + ¢ mulage? ' boy ' (masculine) (singular) mulag- + «a + i mulage 'boys' (masculine) (plural) mulag- + -i + mulagi 'girl' (feminine) (singular) In general, the nouns in Marathi have two forms: (i) .Direet form: This is the same as.the noun stem. (ii) Obligue form:The noun stem is modified when the postposition is attached to it: e.g. (3.1.2) ' sea-15 'to the man man (to) oblique form m’ asE-ca man (of) 'of the man' oblique form In Marathi the numeral gk 'one' is used as an indefi- nite article. e.g. ( .1. ) k " b" " di " 3 3 (oie/a) (gringo) / i333? / (1.3;) The stress on gk in the following sentence, (3.1.4) jan-ni gk pustak anala (John-INSTR) (one) (book) (brought) "John brought one book.‘ 41 gives the reading in which the number of the books is stress- ed (i.e. one and not two, for example), whereas the absence of the stress on gk in the above sentence gives the follow- ing reading: is used and the noun in (3.1.5) 'John brought a book.‘ In Marathi a single formative, which begins with 3; as the third person pronoun, the definite article remote demonstrative,and it agrees with the head number and gender: Masculine to te Feminine ti tya Neuter te ti 'the',?that' 'he' e.g. to mulaga 'the boy', 'that boy' 'the',those','they' e.g. te mulage 'the boys', 'those boys' 'the','that'.'she' e.g. ti mulagi 'the girl','that girl' 'the',those','they' e.g. tyE muli 'the girls','those girls' 'the','that','it' e.g. te pustak 'the book','that book' 'the','those','they' e.g. ti pustaka 'the books','those books' Stockwell et al.(1973, pp.73-74) give the following three types of uses of the definite article in English: (i) Anaphoric (within a sentence). e.g. I saw a cat in the tree this morning, but when I looked this afternoon the cat was gone. (ii) Definite description with relative clause,e.g. 42 T e boy who gave me this book wants it back tomorrow. (iii) Non-linguistically anaphoric, e.g. *3 he moon is full tonight. As in English, in Marathi the definite article is used anaphorically within a sentence, e.g. (3.1.6) an-ni ek pustak finals 'Eni bil-ni John-INSTR one book brought and Bill-INSTR te pustak wacala the book read 'John brought a book and Bill read the book.’ Further, as in English, in Marathi the definite article assigns a definite description to the head NP in a relative clause construction. In this use, the definite article in Marathi is sometimes called the correlative pronoun, (e.g. (3.1.7) jo kal 515 to minus lekhak She REL yesterday came COREL man writer is 'The man who came yesterday is a writer.’ Note that in the above sentence the relative clause precedes the head NP. More will be said about it later. However, in contrast to English, in Marathi the definite article is not used in non-linguistically anaphoric contexts, e.g. (3.1.8) a. pruthwi ; surya ; raStrapati 'the earth ' 'the sun "the president of a nation ' 43 b . pruthwi gol .ahe earth round is 'The earth is round.’ c. raStrapati Ej ithe yenar President today here coming 'ahet lshonorific 'The President is coming here today.’ Note that in Marathi sometimes a noun is not preced- ed by either the definite or the indefinite article. In those cases the definiteness of the noun can be determined only within the context in which a certain sentence is uttered, e.g. harawala lost jbn-ni pustak John-INSTR book (3.1.9) The above sentence may mean either John lost book. .2 2; John lost the book. More examples follow: kutra bhunkalg dog barked 'A/the dog barked.‘ (3.1.10) a. e V - b. Siwha orarala lion roared 'A/the lion roared.’ c. 33n-ni kawita lihili John-INSTR poem wrote 'John wrote a/the poem.’ In English, generally the indefinite as well as the definite article is used in generic sentences, e.g. 44 A/the cat is a mammal. An/the owl can see in the dark. In Marathi, on the other hand, no article is used with generic NPS, e.g. (3.1.11) a. wagh krur prani ahe ti er ferocious animal is 'A the tiger is a ferocious animal.‘ b. manjar paliw prani She cat pet animal is 'A/the cat is a pet animal.’ Descriptive adjectives in Marathi agree in number and gender with the head noun, e.g. (3.1.12) cangalg mulaga , cangali mulagi masculine masculine’ feminine feminine singular singular Singular Singular 'good boy" ’800d (girl‘ Note that when the postposition is attached to the NP, all of its constituents are Modified, e.g. (3.1.13) to n11; ghoré'.’ the/that blue horse 'the/that blue horse' - .- hv—- tya nilya g orya-la the/that blue horse-to 'to the/that blue horse' Word order in Marathi is dominantly SOV. However, due to its inflectional nature, Marathi allows a greater latitude in this regard, e.g. 45 (3.1.14) Ken bil-la kathi-ni marato (=SOV) John Bill-to stick-with hits 'John hits Bill with a stick.‘ The constituents of the above sentence may be scrambled to derive the following variant sentences, which are perfectly grammatical: (3.1.15) a. bil-la 35h kathi-ni marato (=OSV) b. 35h kithi-ni marato bilolE (=svo) c. kathi-ni bil-la marato ESn (=OVS) In the absence of any contrastive stress, all of the above sentences have the Same meaning. In Marathi, yes/no type questions are formed either by adding the interrogative pronoun ka(y) at the end of the sentence or by use of rising intonation, e.g. (3.1.16) a. jen-ni pustak anala John-INSTR book brought 'John brought a/the book.‘ b. jBn-ni pustak anala ka(y)? John-INSTR book brought QUEST 'Did John bring a/the book?‘ 0. an-ni pustak anala? John-INSTR book brought 'Did John bring a/the book?‘ In contrast to English, in Marathi interrogative words such as kog 'who', kay 'what', kuthe 'where', kadhi 'when' are not moved to sentence initial position, e.g. (3.1.17) a. an-ni konala marala? John-INSTR wHom beat 'Who did John beat?’ 46 b. jBn kuthe gelg? John where went 'Where did John go?’ More will be said about question formation in Marathi and its interaction with relative clause formation. The relative clause sentence given in (3.1.7) makes it clear that in Marathi the relative pronoun is formally distinct from the interrogative pronouns. Phrase Structure Rules for Marathi The following phrase structure (PS) rules provide the constituent structures necessary for the application of the transformations proposed for Marathi in this work. (3.1.18) a. s _., sn J NP AUX pop b. NP «o NPn (S) (Det) N Post-P(hrase) c. Post-P —9 NP Postposition d. AUX ~v Tense (M) (Aspect) e. PDP -v (ADV) VP f. VP —v (NP) (Post-P) (Post-P) VB In the PS rule given in (3.1.18) a, Sn accounts for conjoined sentences. As mentioned earlier, it is assumed in this study that restrictive and non-restrictive relative clauses are derived from an underlying set of unordered clauses. 1.7 The Internal Structure of the NP in Marathi: In the PS rule given in (3.1.18) b. NPn accounts for the following sentence with a conjoined NP: (3.1.19) 35h. ani meri sarakhe disatat John and Mary similar look 'John and Mary look similar.‘ Note that the conjoined NP Hen'agi meri 'John and Mary' in the above sentence cannot be transformationally derived: (3.1.20) *jan s§rakh§' disate 'Eni meri sgrakhi disate 'John looks Similar and Mary looks similar.’ Conjunction V - . . - h . - reduction -+ jan ani meri sarak e disatat The phrasal conjunction in the above sentence will be derived in the base structure by PS rule (3.1.18) b, which expands NP into NPn . Furthermore, this PS rule shows that NP contains N as the obligatory constituent. As mentioned earlier in this section, in Marathi nouns may occur with or without articles: (3.1-21) ek a minus / amba- / :1in to man mango lamp the Also, NPs in generic sentences occur without articles (see (3.1.11) ). Further, PS rule (3.1.18) b says that N may Optionally be preceded by the constituent Det which includes articles, demonstratives, numerals, and quantifiers. The relative 48 order of these constituents in Marathi is as follows: (3.1.22) Det-rDemon- A Posses-A Article n{h1umera];5[escrip-n t strative Sive Adj Quanti 1ve Adj fier (3.1.23) a. h§ jBn-ca sadara this John's shirt b. te Kan—ce don pandhare surekh kutre those John's two white beautiful dogs The expansion of NP into S(Det)N represents the complement structure in Marathi as shown in (3.1.24) a and b: (3.1.2u) a. A /NP\ Aux VIP S__ N VB jan-ni pustak harawala he k ara John-INSTR book lost it true +N +VB +pron -V b. /SI\ NP Aux VF S Det N NE A jan-ni pustak hi gosta khari‘ John-INSTR book this thing true harawala lost [ +VB] The complementizer k; 'that' appears on the surface only if the embedded complement sentence is extraposed, e.g. 49 (3.1.25) he khara She ki Ken-n1 pustak harawala it true is that John-INSTR book lost 'It is true that John lost a/the book.’ In Marathi the complement clause generally precedes the main clause, in fact most subordinate clauses do. PS rule (3.1.18 b)says that NP may also be expanded into a Postpositional phrase, whose internal structure is NP + Postposition. Most of the postpositions in Marathi are bound formatives. In some cases Post-PS occur in the subject position in Marathi sentences, e.g. (3.1.26) a. xii meri-la ek harin disala yesterday Mary-to a deer see-PAST 'Yesterday Mary saw a deer.’ b. jhn-la' Zmbi awarato John-to mango likes 'John likes a mango.’ Sentences (3.1.26) a and b are instances of the so-called dative construction in Marathi: in this construction the subject is in the dative case and the main verb (which is one of the verbs of perception) agrees with the object. Another instance in which the subject NP is followed by a postposition is the so-called ergative construction in Marathi. In this type of construction, in the past tense the subject of a transitive clause takes the postposition -ni (i.e. Instrumental case), and the verb agrees with the object in the sentence: (3.1.27) XSn-ni 'amsa khalla , . John-INSTR mango ate John ate a/the mango. 50 3.2 A Restrictive relative clause in Marathi Colin Masica (1972, p.»198) says the following about the relative clause construction in Indo—Aryan languages. The typical INDO-ARYAN construction is the correla- tive construction-- a subvariety of relative pronominal construction. The sentence to be relativized is inserted to the left of the main sentence. The relative pronoun- adjective, which begins with j- (and is therefore distinct from the corresponding interrogative which generally begins with k-: this is incidentally precisely the Same distinction, inherited from Sanskrit by modern Indic languages, whose loss in later Slavic is described elsewhere in this volume),is then attached to the left of the relativized noun, while the correlative pronoun-adjective, which is usually the same as the remote demonstrative, is simultaneously attached to the left of its counterpart in the main sentence. In Marathi the relative clause construction may occur in one of the following surface structure configurations: . 1 Prospective relative (pro-relative): the relative clause (Srel) precedes the head noun phrase (NPhead)' (3.2.1) jo minus kZi 315 to mahus lekhak'ahe REL man yesterday came COREL man , writer is 'The man who came yesterday is a writer.‘ Retrospect1ve relative (retro—relative): Srel follows Nphead. (3.2.2) to manus jo(m§r}us) 1:21 31; to COREL man REL man yesterday came COREL ('manus) lekhak ahe man writer is 'The man who came yesterday is a writer.’ Adjoined relative (Ad-relative): NPs and Srel do not form a constituent: Srel is merely subordinate to the clause con— taining Nphead' 51 There are two types of adjoined relatives: a. Preposed relative: in this type, the relative clause precedes the main clause. (3.2.3) jbn-ni 3e pustak 'Enala mi te John-Erga- REL book brought I COREL tive (pustak) wacala book readpast 'I read the book which John brought.’ b. Extraposed relative: in this type, the relative clause follows the main clause. (3.2.4) mi te pustak wacala jam-hi I COREL book read John-Ergative 3e (pustak) anala REL book brought 'I read the book which John brought.’ Next, consider the following sentence: (3.2.5) jo m'a'qus 1:31 315 to means lekhak She REL man yesterday came COREL man writer is 'The man who came yesterday is a writer.’ In the above sentence 1; is a relative pronoun and p: is the so-called correlative pronoun. Both of them agree in number, gender, and case with the nouns they precede. .The other variants of sentence (3.2.5) are: (3.2.6) a. jo mgnus 221 ala to lek 52 - - hak Ehe b. jo ha; 515 to sagas lekhak 'Ehe c. 4’ .. n . .. n n d. to minus jo kal 31; to lekhak She e. to ¢ " " " to m'a-nus lekhak ahe f . to ¢ " manus 1:31 213 to lekhak She g. to manus lekhak She jo £51 315 (to minus) Clearly the above sentences involve such syntactic operations as copying and deletion of constituents, which are discussed later in this chapter. There are the following surface structure differences between the relative clause construction in Marathi and its counterpart in English. (1) (ii) (iii) (iv) In Marathi generally the relative clause precedes the head NP, whereas in English it always follows the head NP. Unlike in English, in Marathi in the relative clause construction there is a correlative pronoun t-- which precedes the head NP. In Marathi the head NP may be deleted, whereas in English the head NP is never deleted. In English, but not in Marathi, the relativized NP is always deleted. Note that sentence (3.2.5) may be translated into 53 English as either (3.2.7) a or b. (3.2.?) a. The man who came yesterday is a writer. b. A man who came yesterday is a writer. The reason for this is that as mentioned earlier in this chapter, in Marathi if the noun is not preceded by an article, its definiteness can be determined only within the context of the nature and extent of the hearer's knowledge of the speech act in question. Note that in sentences (3.2.7) a and b the speaker has a certain individual in mind. That is, for him there exists the (same) referent for the head NPS in those sentences. However, for the hearer there may exist the referent for the head NP in sentence (3.2.7)a only. In this sentence the speaker assumes that the hearer knows about the man who came yesterday, hence he (i.e. the speaker) chooses the definite NP, the man. 0n the other hand, in sentence (3.2.7) b the speaker assumes that the hearer does not know about the man who cameyyesterday, hence he uses the _indefinite NP, a man. Thus the use of a definite/ indefinite NP in the relative clause construction in English serves as an aid to the hearer in the identification of the referent of the head NP. For example, in sentence (3.2.7) a the speaker through the use of the definite NP, seems to reassure the hearer that in his (i.e. the hearer's) world there exists the referent for the NP that the restricting sentence (i.e. who came yesterday) is true of. In sentence (3.2.7) b, on the other 54 hand, the speaker, through the use of an indefinite NP .seems to assume that there may not exist in the hearer's world the referent for the NP. Consider the following frag- ment of a discourse: A: A.man who came yesterday is a writer. B: I know him. I have read some of his work. As the above discourse Shows, the hearer may have the referent for an indefinite head NP. Sentence (3.2.7) a shows that in contrast to English, in Marathi the speaker may not overtly suggest what he thinks to be the extent of the hearer's knowledge about the referent of the NP. Notice that the non-definiteness of the head NP may be overtly indicated in sentence (3.2.7) a .e.g. (3.2.8) jo kal 315 to ek manus lekhak {She REL yesterday came COREL a man writer is 'A man who came yesterday is a writer.‘ However, the definite article cannot be used in that sentence to indicate the definiteness of the head NP, as exemplified in the following sentence: (3.2.9) * jo 1651 :1; to to mantis lekhak zhe REL yesterday came COREL the man writer is In order to understand why the above sentence is ungammatical, it is necessary to figure out the precise function of the so-called correlative pronoun in Marathi. Masica (1972. p. 198) correctly observes that the correlative pronoun "is 55 usually the same as the remote demonstrative.” It can be argued however, that the remote demonstrative is used as a correlative pronoun in its neutral sense. S. Y. Kuroda (1969, p. 271) says: . . . the word ppgp may also be used sometimes in a neutral way with respect to this-that Opposition. In this use that is much like ii, as a pronoun and the, as a deter- miner. Thus in sentence (3.2.8), reproduced below, (3.2.8) jo 1.51 513 to ek m'a'nus lekhak 'ahe REL yesterday came that one man writer is the remote demonstrative ppgp relates the head NP ek mEpus to the preceding relative clause jO kal El; to. Notice that in English the relative clause generally follows the head NP, hence the relative pronoun can relate the head NP to the following relative clause, e.g. The girl who lives upstairs plays the sitar. 0n the other hand, in Marathi. due to such marked tendencies 'as the preposed relative clauses and SOV word order, the relative pronoun cannot relate the head NP to the preceding relative clause. Therefore in the absence of the remote demonstrative pp in the following sentence (3.2.10) *jo kal 513 ek minus lekhak 'ahe it is difficult to perceive that the two underlined clauses are related to each other. What is suggested here 56 is that in the parsing of the above sentence, the presence of the remote demonstrative serves as the surface structure clue that enables the hearer to relate the head NP to the preceding relative clause. In this sense the remote demon- strative complements the relative pronoun. Furthermore, as noted by Kuroda (1969, p. 271). it is "much like ip, as a pronoun and ppg as a determiner". For example, in sentence (3.2.8) the remote demonstrative pp, in a certain sense, assigns some degree of definiteness to the non-definite head NP ek mépus: 'a man', namely, 'that one man'. This fact probably explains why sentence (3.2.9) is ungrammatical: since the remote demonstrative :3 already denotes (some degree of) definiteness for the head NP, the overt presence of the definite article pp is redundant. It shouLd be mentioned here that in some cases the adjective §§_ 'such: of this kind' is used in place of the remote demonstrative in relative clause sentences: and the choice of one or the other seems to be governed by a certain presupposition on the part of the speaker. For example, consider sentences (3.2.11) a and b: (3.2.11) a. Xi paisewali ahe tya ekE 651-§1 REL rich is that a woman-with .00. v V. .VVh-n :— lagna kara1c1 an-ci 1cc a she to marry John's wish is 'John would like to marry a woman who is rich.‘ 57 b. 3’1 paisewali an. - a‘s’E ek; 1:31-31 REL rich is such: a woman-with of this kind lagna kar'aig i gen—‘61 1361‘s.: Ehe to marry John's wish is 'John would like to marry a woman who is rich.‘ Notice that for both the sentences in Marathi there is only one English translation. Furthermore, note that the English translation is ambiguous: in one case John has a particular woman in mind to marry and the relative clause who is rich merely provides descriptive information about that woman: let us call this reading (1). In the other case, John does not have a particular woman in mind: it is possible that no such woman exists: let us call this reading (ii). The English translation of sentences (3.2.11) a and b is an example of opaque contexts (Quine, 1960). Notice that unlike their English counterparts, sentences (3.2.11) a and b in Marathi are not ambiguous: sentence (3.2.11) a corresponds to read- ing (i) and sentence (3.2.11) b corresponds to reading (ii). The only difference between the two sentences is that sentence (3.2.11) a contains the remote demonstrative tyi 'that', whereas sentence (3.2.11)b contains the adjective §§§.'such: of this kind'. The presence of the remote demon- strative in sentence (3.2.11) a expresses the presupposition, namely that there exists the referent for the NP ek bai 'a woman', but the word géé 'such: of this kind' in sentence (3.2.11) b does not convey any such presupposition. However, note that in the following two sentences the above difference between the remote demonstrative and the adjective agg'is 58 neutralized: " up -V (3.2.12) a. Xi paisewali asel tya bai-si REL rich may be that woman-with an-IE lagna karaigi iggha’ ahe John-to to marry wish is 'John would like to marry any woman who is rich.‘ b. Xi paisewali asel .aSE' bEi-Si REL rich may be such: woman-with of this kind — —V - — jam-la lagna karaici iggha ahe John-to to marry wish is 'John would like to marry any woman who is rich.’ As their English translation shows, sentences (3.2.12) a and b have the same meaning. There is no presupposition in either of these sentences that there exists a particular woman of whom the restricting sentence is true. Another interesting thing to note here is that the sense of the quantifier is expressed through the verb aggl. Note that the substitution of _a_‘s'__a._ 'such: of this kind' for 39 'that' in sentence (3.2.5) results in an ungrammatical sentence (3.2.13). (3.2.5) jo £51 515' to ek manus lekhak She REL yesterday came that a man writer is 'A man who came yesterday is a writer.‘ (3.2.13) * jo 1651 3'13 as; ek masts lekhak Ehe such: of this kind Sentence (3.2.5) presupposes that there exists a certain man of whom the relative clause jo kal'alz’is true. The ungra- mmaticality of sentence (3.2.13), therefore, Shows a 0 v- 0 I constraint on the use of asa as a correlative word in the 59 relative clause construction, namely that it cannot be used if the referent for the head NP is presupposed to exist.2 Kuroda (1969, p. 271) points out that the remote demonstrative ppgp in its neutral sense " is much like i3, as a pronoun." Sentence (3.2.6) a, one of the variants of the relative clause sentence, shows that this is indeed the case in Marathi. (3.2.6) a. jO m‘a'nus kal '51; to lekhak ahe REL man yesterday came that one writer is i.e. he 'The man who came yesterday is a writer.‘ In the above sentence the remote demonstrative takes on the function of the personal pronoun 'he' since the head NP is deleted. (As mentioned earlier, in Marathi the definite article, the remote demonstrative and the third person pronouns are morphologically identical.) 3.3 Formation of a restrictive relative clause in Marathi: This section deals with the various syntactic opera- tions that are necessary to derive restrictive relative clauses in Marathi: RELATIVE CLAUSE FORMATION (RCF) As mentioned earlier, it is assumed in this study that restrictive and non-restrictive relative clause sentences are derived from an underlying set of clauses. Given a set Of unordered clauses in the underlying 60 structure, the speaker decides which one is to be the subordinate clause (i.e. the relative clause) and which one is to be the main clause. Generally the following considera- tions seem to determine this choice: (i) ng_information vs. new:ipformation: in a restrictive relative clause, the speaker provides the hearer with what he believes to be the Old information (i.e. presupposition). so that on the basis of that information the hearer can pick out the referent for the head NP: in the main clause the speaker introduces what he sincerely believes to be the new information (i.e. assertion). Also, as discussed earlier in this chapter, in the case of restrictive relative clauses in‘ English, the speaker uses a definite NP (e.g., sentence (3.2.?) a) as an aid to enable the hearer to pick out some individual. (ii) Primary vs. secondary_information: in a non-: restrictive relative clause, the speaker assumes that the referent of the head NP may or may not be known to the hearer. Therefore his choice of the subordinate clause and the main clause correlates with what he sincerely believes to be the primary (i.e. relevant) information as Opposed to the secondary (i.e. non-relevant) information. Thus he intro- duces the secondary information, which may or may not be new to the hearer, in the subordinate clause and the primary information in the main clause. In the linguistic literature it has been noted (e.g., Bolinger (1965), Firbas (1957)) that Old information generally 61 appears towards the beginning of sentences and new informa- tion towards the end. There is nothing surprising about this: in fact it is quite natural to expect this in a natural lan- guage whose primary function is communication. It would be infelicitous on the part of the speaker to introduce new information prior to the necessary background information.3 The surface order of the relative clause and the main clause in Marathi is in keeping with the above communicative postulate in that the (restrictive) relative clause (which contains the background information) precedes the main clause (which contains the new information). Returning to the formulation of RCF in Marathi, at the time of the application of this rule, the following is the order of the clauses: £31 I. J . S . main ma1n At this point the following question arises: When is the remote demonstrative inserted in the relative clause construc- tion in Marathi? Masica (1972) suggests that the relative pronoun and the correlative pronoun.(i.e. the remote demon- strative) are inserted simultaneously. In other words, according to him, RCF inserts both of them. One may argue that RCF inserts only the relative pronoun, while the corre- lative arises anaphorically» In the present study, I assume the latter proposal, namely that'RCF inserts only the erelative pronoun and the remote demonstrative arises , 62 anaphorically. One of the reasons for this choice is that the anaphoric definitization process is needed elsewhere in the language on independent grounds: Kuroda (1968) and Postal (1966) argue that the first step in pronominalization of personal pronouns is definitization. In Marathi the pronominalization process involves the following steps: (3.3.1) 0 O O O O emu1ag§i O O O O O omLIlagai 0 O O 0 boy boy ANAmORIC DEFINITIZATION O O O O O O 0 O n O O O O O O 0:: O O O u 0 O O O O O O O TRANSFER OF NOUN FEATURES (phonological rules will spell out the shape of the inflectional ending 21112) OOOOOOOO..OOOOOCO-t_-o- 0.0"00000000 NPDELETION 0.00.00...0000000‘?_Q IOO¢OIOOOOOO Since the above processes are required to derive relative clause sentences, they will be formulated presently. First let us consider the formation of RCF: 63 (3.3.2) RCF W NPi X ] NPi S -S SD 1 2 3 4 SC 1 j + 2 3 4 Condition: ; does not contain any NP coreferential with 2 of SD (3.3.3) ANAPHORIC DEFINITIZATION: X NP. Y NP. Z l 1 SD 1 2 3 4 5 =5 30 1 2 3 t- +4 5 Notice that the way in which the above transformation is formulated, it can be applied either within or across a sentence boundary, whereas RCF can be applied only across -a sentence boundary. TRANSFER OF NOUN FEATURES: Note that RCF attaches j; to the relativized NP and ANAPHO- RIC DEFINITIZATION attaches p; to the head NP. As mentioned earlier, the relative pronoun and the correlative pronoun agree in number, gender and case with the nouns they precede. Therefore we require a transformation which copies out the relevant features of N onto both 1; and p; : 64 (3.3.4) TRANSFER OF NOUN FEATURES: P - ' _. a plural X E: Y Bmasculin Z [feminine N 1‘1 ”NP NP SD 1 2 3 4 5 = so 1 2 3 4 5 .Lplural B masculine r feminine Note that since Marathi has three genders (i.e. masculine, feminine and neuter), the third gender can be minimally determined in terms of the feature specification for any two genders, e.g. + masculine - masculine - masculine -feminine + feminine - feminine The way in which the above transformation is formulated, it can copy out features from N onto 3; as well as j;. Of the three rules formulated above, RCF and ANAPHORIC DEFINITIZATION need not be ordered with respect to each other. However, as previously mentioned, ANAPHORIC DEFINITI- ZATION may apply either within or across the sentence boun— dary. Therefore, ANAPHORIC DEFINITIZATION may be said to ~precede RCF. TRANSFER OF NOUN FEATURES must follow both ANAPHORIC DEFINITIZATION and RCF. 65 Given the above three transformations, sentence (3.2.5), reproduced below, may be derived from the (simpli- fied) underlying structure given in (3.3.5) as follows: (3.2.5) jo minus kgl £15 to manus lekhak 'Ehe REL man yesterday came that man writer is (3.3.5) (manus kal ala) (manus lekhak She) ANAPHORIC DEFINITIZATION ” " " E: u u u RCF i:- " fl " 9' 00 n n TRANSFER OF NOUN FEATURES jg .. .. .. £2 .. .. .. In order to derive the relative clause sentences given in (3.2.6) a through (3.2.6) g, which are variants of sentence (3.2.5), the following transformations are required: NP DELETION: Consider sentences (3.2.6) a and b, reproduced below: (3.2.6) a. jo manus 1651 '51; to lekhak 'ahe REL man yesterday came he writer is b. jo kEl 315' to minus lekhak 'ahe REL yesterday came that man writer is In sentence (3.2.6) a, the head NP is deleted. whereas in sentence (3.2.6)b, the relativized NP is deleted. The follow- ing transformation may be given to account for the deletion 66 of the head NP as well as of the relativized NP in the above sentences: (3.3.6) NP DELETION: X {t-} NP Y 3.. SD 1 2 3 4 so 1 2 (p L: : Notice that since the above transformation may delete either one or both of the NPS in sentence (3.2.5) the following sentence results: (3.3.7) 30 1:21 312 to lekhak Zhe REL yesterday came he writer is The above sentence is perfectly grammatical. It is possible to partially recover the deleted information on the basis of the inflected categories in the sentence: that is. the inflectional endings of the verb gig; and that of the rela- tive pronoun as well as the correlative pronoun in sentence (3.3.7) enable us to know that the deleted NPS must be masculine, third person. Singular. etc. RELATIVE PRONOUN DELETION: Consider sentence (3.2.6) o, reproduced below: (3.2.6) o. R31 'SIE’ to manus lekhak ‘ahe yesterday came that man writer is In the above sentence the relative pronoun jg is deleted. This deletion rule may be given as follows: 6? (3.3.8) RELATIVE PRONOUN DELETION: x j- Y [+REL] so 1 2 3 III-‘D so 1 T 3 Condition: 3 does not contain NP with the correlative pronoun. LEFT DISLOCATION: Consider sentences (3.2.6) b and d, reproduced below: (3.2.6)b. jo kal 315 to manus lekhak (ahe REL yesterday came that man writer is d. to minus jo £51 '31; to lekhak ‘Ehe that man REL yesterday came he writer is In sentence (3.2.6) d, the head NP along with the remote demonstrative is moved to the front of the sentence, leaving behind a copy. The constituent(s) thus moved becomes the topic of the sentence. Following this movement, NP DELETION deletes the head NP from its original position, and the remote demon- strative assumes the function of the personal pronoun pp 'he'. The rule which moves the head NP along with the remote demonstrative and leaves a copy may be called LEFT DISLOCATION: X [ t- NP Y] Z NP NP SD 1 2 3 II‘D SC 2# 1 2 3 (Where # is a sentence boundary.) 68 EXTRAPOSITION: Consider sentence (3.2.6) g, reproduced below: (3.2.6) g. to minus lekhak ahe jo 1:31 513 that man writer is REL yesterday came ( to manus ) that man In the above sentence the relative clause is extraposed to the end of the sentence. Also, optionally, the head NP along with the remote demonstrative is moved to the end of the sentence. These two rules, EXTRAPOSITION and RIGHT DISLOCA- TION, may be given as follows: (3.3.10) EXTRAPOSITION: W '- NP. X] [Y t- NP. 2] [s J 1 (s s 1 '8 SD 1 2 SC d? 2+1 u-fi It may be added that, contrary to p0pular belief. EXTRAPOSITION in English is not always Optional. As mention- ed earlier (Ziv and Cole.1974), extraposed relative clauses do not have quite the meaning as non-extraposed relative clauses. Furthermore, the following two Sentences (Borkin, 1972) do not have the same meaning. (3.3.11) a. 'The old woman who is rich will marry the handsome young man.‘ b. 'The old woman will marry the handsome young man who is rich.’ 69 (3.3.12) RIGHT DISLOCATION: X [N t- NP] Y P NP SD 1 2 3 SC 1 2 3 # 2 (Where # is a sentence boundary.) The relative clause sentences (3.2.5) and (3.2.6) a through g;are derived from their common underlying structure by applying the transformations formulated above, as follows: ANAPHORIC DEFINITIZATION RCF and TRANSFER OF NOUN FEATURES —"’ (3.2.5) jo m'a'nus 1:51 315. to REL man yesterday came that minus lekhak ahe man writer is (i.e. preposed relative) ANAPHORIC DEFINITIZATION RCF TRANSFER OF NOUN FEATURES NP DELETION —e (3.2.6) a. jo mE'nus 1:51 315 to REL man yesterday came that one lekhak ‘ahe (i.e. preposed writer is relative) =======:h b- 3° ANAPHORIC DEFINITIZATION RCF TRANSFER OF NOUN FEATURES RELATIVE PRONOUN DELETION ‘====’ (3.2.6) C. ANAPHORIC DEFINITIZATION RCF TRANSFER OF NOUN FEATURES LEFT DISLOCATION NP DELETION ——Ih (3.2.6) (1. 70 1:21 31; to m-a—nus REL yesterday came that man lekhakizhe writer is kal 21; to manus yesterday came that man lekhak .ahe writer is (i.e. preposed relative) to manus jO kal 515 that man REL yesterday came to lekhak ahe that one writer is to jo 1.31 315 that one REL yesterday came to minus lekhak'ahe that man writer is to jo 1:1 515’ that one REL yesterday came to lekhak ahe that one writer is (i.e. retrospective relative) 71 ANAPHORIC DEFINITIZATION RCF TRANSFER OF NOUN FEATURES EXTRAPOSITION (RIGHT DISLOCATION) NP DELETION —_> (3.2.6) g. to manus lekhak ahe jo that man writer is REL 1:31 31; ( to manus ) yesterday came that man (i.e. extraposed relative) The derivation of sentences (3.2.5) and (3.2.6) a through g shows that the transformational rules mentioned above may be applied with the following orderings. (3.3.13) ANAPHORIC DEFINITIZATION RCF TRANSFER OF NOUN FEATURES NP DELETION C RELATIVE PRONOUN DELETION ///’T§ LEFT DISLOCATION EXTRAPOSITION C RIGHT DISLOCATI ON Most of the above transformations are independently motivated in that they are needed elsewhere in the language For example, the pronominalization process in Marathi requires ANAPHORIC DEFINITIZATION. TRANSFER OF NOUN FEATURES. and NP DELETION. LEFT DISLOCATION is employed in the 72 tOpicalization process, e.g. (3.3.14) j’an. meri-la toi Ewa‘II‘at n'ahi Johfi Mary-to he like not 'John, Mary does not like him.’ EXTRAPOSITION is needed to extrapose embedded complement sentences (see sentence (3.1.25)). 3.4 ,A non-restrictive relative clause in Marathi There is a discussion in section 3 of chapter 2 regarding the underlying structure of non-restrictive relative clauses in English. As mentioned earlier, in the present study the Conjunction analysis is assumed for restrictive and non-restrictive relative clauses in Marathi. Furthermore, there is a discussion in section 3 of this chapter concern- ing the speaker's choice of the subordinate clause vs. the main clause in non-restrictive relative clauses. Consider the following non-restrictive relative clause sentences in English: (3.4.1) a. The old man, whom we ,met, knew the way. b. Bill, whom we met, knew the way. In these sentences the speaker assumes that the individuals referred to in the head NPS are already known to the hearer. The information represented in the relative clause may be either new or old, and the information provided in the main clause is assumed to be new. Intuitively it is clear that although the definite NP, the man, and the proper name, 73 Bill, in the above sentences refer to already identified individuals, the prOper name is more definite than the definite NP. It was pointed out earlier that in the case of restrictive relative vs. indefinite NP is nature and extent of the hearer's knowledge. clauses the speaker's choice of definite the result of his assumption about the From what is said above, restrictive and non-restric- tive relative clause sentences stand in the on a continuum of identifiability of a referent. following relation 5 Degree of identifying function of relative clauses High r-— r-- J" Low (1) Restrictive relative clauses with (2) (3) (4) (5) indefinite head NPs Restrictive relative clauses with definite head NPS Non-restrictive relative indefinite head NPS Non-restrictive relative definite head NPS Non-restrictive relative proper names as head NPS clauses with clauses with clauses with In English there is a formal difference between restric- tive and non-restrictive relative clauses, in that the latter, but not the former, are marked by intonation commas. In Marathi the difference between these two types Of clauses is sometimes conveyed through the respective order Of the relative clause 74 and the head NP (Kelkar,1973). Consider the following sentences: (3.4.2) a. jo mumbait rahato to maza bhiu REL Bombay-in lives that my brother wakil She lawyer is 'My brother who lives in Bombay is a lawyer.’ b. maza bhau jO mumbait rahgto (to) my brother REL Bombay-in lives that one/he wakil ahe lawyer is 'My brother, who lives in Bombay, is a lawyer.' AS the English translation Shows, sentence (3.4.2) a is a restrictive relative clause construction, whereas sentence (3.4.2)b is a non-restrictive relative clause construction. Note that these two sentences are identical, except for the respective order of the relative clause and the head NP: in sentence (3.4.2) a the relative clause precedes the head NP (i.e. the preposed relative), whereas in sentence (3.4.2) b it follows the head NP (i.e. the retrospective relative). (Incidentally, it may be mentioned here that Junghare (1973) claims that in contrast to restrictive relative clauses, in non-restrictive relative clauses in Marathi the definite determiner pp is deleted. However, note that in sentence (3.4.2) b, pp may in fact Optionally appear and yet we get the non-restrictive reading.) However, as the following example Shows, there is something more than just the respective order of the relative clause and the head NP that distin- guishes the two types of relative clauses: (3.4.3) 75 to maz; bhau jo mumbait rahato that my brother REL Bombay-in lives to wakil ahe that one/he lawyer is 'My brother who lives in Bombay is a lawyer.‘ In the above sentence. although the relative clause follows the head NP, we get the restrictive reading. This is because the head NP is preceded by the remote demonstrative pp which determines the scope of that head NP (Junghare,1973). And Since it is quite possible for someone to have more than one brother, we get the restrictive reading. From the above discussion the following hypothesis may be proposed regarding the formal difference between restrictive and non-restrictive relative clauses in Marathi: (3.4.4) a.... In restrictive relative clauses the head NP (1) is alwayg preceded by the remote demonstrative t- which anaphorically refers back to its head. (ii) may either pgggggg or QQLLQE the relative clause (i.e. retro-relative or pro-relative). b.... In non-restrictive relative clauses the head NP (i) is ggyg; preceded by the remote demonstrative E: (ii) alwayg pggggggg the relative clause (i.e.retro- relative). The above hypothesis predicts that the (a) sentences 76 below yield non-restrictive readings and (b) and (c) sentences, restrictive readings. That is indeed the case: (3.4.5) a. maze waPil je mumbait rahatat te my father REL Bombay-in lives that one Honorific H H H wakil Ehet lawyer isH 'My father, who lives in Bombay, is a lawyer.' - v b.*te maze waril 3e mumbait rahatat that one/heH myH father RELH Bombay-in livesH te wakil (ahet that one/heH lawyer isH *‘One of my fathers who lives in Bombay is a lawyer.‘ c.-- - .V c.*je mumbait rahatat te maze waril RELH Bombay-in livesH that one/heH myH father wakil Shet lawyer isH *‘One of my fathers who lives in Bombay is a lawyer.’ V -’ —. (3.4.6) a. XSn jo sikagot rahato to John REL Chicago-in lives that one/he aj ithe yedar ‘Ehe today here coming is 'John, who lives in Chicago, is coming here today.’ b. to XSn jo SikEgot rahato to that one/he John REL Chicago-in lives that one/ he -a.j ithe yenzr ahe today here coming is ‘The John who lives in Chicago is coming here today.’ 77 v - - - ‘_ c. jO sikagot rahato to 33h aj REL Chicago-in lives that one/he John today ithe yen'a'r ahe here coming 13' 'The John who lives in Chicago is coming here today.’ Note that sentences (3.4.5) b and c are ungrammatical for extralinguistic (i.e. culture-Specific) reasons. However, one can find several counterexamples to part b of the hypothesis given in (3.4.4). Many occur in rhetorical speech and writing, e.g. (3.4.7) jglnni gathg. racali te tukaram maharaj REL-INSTRH Gatha composed thatH St. Tukaram dehula’ janmale Dehu-at was bornH 'St. Tukaram, who composed the Gatha, was born in Dehu.’ (3.4.8) gandhi-nni jzznna guru manale te Gandhi-INSTRH REL-toH teacher regardedchatH gokbale mawal hote Gokhale moderate wasH 'Gokhale, whom Gandhi regarded as (his) teacher, was a moderate.‘ Note that in both sentences, which are non-restrictive relative clause constructions, the remote demonstrative precedes the head NP which, in turn, follows the relative clause. In all the non-restrictive relative clause sentences 78 mentioned above, the non-restrictive clause merely adds some information or comment about the NP whose referent is already known to the hearer. In other words, the non- restrictive relative clause can be removed from the main clause and yet the primary message can be communicated. However, in some cases the non-restrictive relative clause does seem to form an integral part of the main clause. Consider the following sentences: (3.4.9) a. joSi 3e ithe rahatat te saikal-ni Joshi RELH here livesH that one/heH bike-with j'a'tEt goesH '(Mr) Joshi, who lives here, rides a bike.’ .V. v V- - b. 3031 je sattar warsace ahet te Joshi RELH seventy years-o H isH that one/heH ' sZikal-ni 33.5“: bike-with goesH '(Mr) Joshi, who is seventy years Old, rides a bike.’ (3.4.10) a. 35h jo warati rahato to freng John REL upstairs lives that one/he French cangala bolto well Speaks 'John, who lives upstairs, Speaks French well.’ . . - . V - . -' ‘- b. 33n. jo pzr1s-madhe kahi warsa rahila hota John REL Paris-in some years lived to freng' cangala bolto that one/he French well speaks 'John, who lived in Paris for some years, speaks French well.’ 79 Note that in sentences a above the relative clause provides some additional information about the head NP (i.e. the topic), whereas in sentences b, the information provided in the relative clause relates not only to the tOpic but also to the comment in the main clause: in sentence (3.4.9) b the relative clause does not merely provide the additional information about the person's age: it emphasizes the fact that the person rides a bike in spite of his age. In sentence (3.4.10) b, the relative clause describes the causal relationship between John's speaking French well and his having lived in Paris for some years. Sentences given in (2.3.13) a and (2.3.14) a in chapter 2. provide more examples of this sort. 7 From the preceding discussion the following remarks can be made about non-restrictive relative clauses in Marathi: (i) In Marathi, non-restrictive relative clauses are not always formally distinct from restrictive relative clauses. Sometimes a distinction is made between them, as shown in (3.4.4) a and b. Thus Marathi is half-way between English, in which these two types of clauses are always formally distinguished, and Japanese (Kuno,1973) in which no such distinction is made. Therefore, in the absence of any formal distinction between restrictive and non-restric- tive relative clauses, their semantic interpretation is determined by the nature and extent of the hearer's 8O knowledge about the referent of the head NP. (ii) Some, if not all, of the non-restrictive relative clauses relate not only to the topic but also to the comment in the main clause. The following transformations in their given order are required to derive a non-restrictive relative clause sentence from its underlying structure: ANAPHORIC DEFINITIZATION RCF TRANSFER OF NOUN FEATURES NP PREPOSING (See below) ANAPHORIC DEFINITIZER DELETION (See below) NP DELETION (3.4.11) NP PREPOSING: x t- NP SD 1 2 3 SC 3 # 1 2 q: 4 : Note that in the case of restrictive relative clauses, LEFT DISLOCATION moves the head NP along with the anaphoric determiner, leaving behind a COpy. On the other hand, in the case of non-restrictive relative clauses, NP PREPOSING moves only the head NP, without leaving its copy behind. It is argued here that this difference is semantically and 81 syntactically well motivated: in the case of the restrictive relative clause, the anaphoric determiner relates the relative clause to its head NP. Therefore, whenever the head NP is moved, the preceding determiner must be also moved along with it. Secondly, in the case of the restrictive relative clause, as mentioned earlier, the anaphoric deter- miner cannot be deleted. For exampleoin sentence (3.2.6) b and d, reproduced below, (3.2.6) b. jo 1:21 21; 1:2 menus lekhak ahe REL yesterday came that man writer is 'The man who came yesterday is a writer.‘ d. t_o_ manus jo 1:21 212 _t_g lekhak She that man REL yesterday came that writer is one ‘The man who came yesterday is a writer.‘ none of the underlined determiners can be deleted. In sen- tence (3.2.6) d, LEFT DISLOCATION has moved the head NP along with the determiner to the front of the sentence. The presence of the determiner pp_before the word lekhak cannot be explained if we treat LEFT DISLOCATION as a chopping rule(Rossl967). However, if we treat it as a copying rule the presence of the determiner pp before the word lekhak can be explained as follows: ANAPHORIC DEFINITIZATION attaches _t__-_ to'the head NP. Since LEFT DISLOCATION is a copying rule, it leaves behind the copy of the determiner as well as the head NP. Therefore. the determiner is not left dangling: it can hang on to the 82 NP node (even after N is Optionally deleted). 0n the other hand, in the case of non-restrictive relative clauses, the referent of the head NP is already known. Therefore the anaphoric determiner is not required in this case. When NP PREPOSING moves the head NP to the front of the sentence, it chops the NP node, hence the anaphoric determiner is left dangling: That explains why it can be (Optionally) deleted. (3.4.12) ANAPHORIC DEFINITIZER DELETION: This rule is applied only if NP PREPOSING has preceded it. X t- Y SD 1 2 3 a so 1 (p 3 The following is a sample derivation of (the restrictive relative clause) sentence (3.4.5) a from its underlying structure in (3.4.13): - V -n— (3.4.13) ( maze waril mumbait rahatat ) myH father Bombay-in livesH - V. . - - ( maze waril wakil ahet ) my father lawyer is H H 'My father lives in Bombay. My father is a lawyer.‘ 83 ANAPHORIC DEFINITIZATION, RCF and TRANSFER OF NOUN FEATURES V .- V. . ______. je maze ,waril mumbait rahatat REL my father Bombay-in lives H H H a. V - te maze waril wakil ahet that my father lawyer is H H H v ... NP PREPOSING ——-§ maze waril Xe maze wagil myH father RELH myH father munbait rahatat te- wakil ahet Bombay-in livesH thatH lawyer isH ANAPHORIC DEFINITIZER DELETION :— V n V -——D' maze waril Xe maze waril myH father RELH myH father mumbait rahatat wakil ahet Bombay-in livesH lawyer isH — v NP DELETION _______.. maze waril is mumbait . myH father RELH Bombay-in rahgtat wakil ‘Ehet livesH lawyer isH 3.5 Survey of two previous studies on relative clauses in Marathi: In this section I take a brief overview of the two . . 8 . prev1ous studies on relative clauses in Marathi: Junghare (1973) and Kelkar(1973). Junghare (1973, page 251) sets out "to clarify the confusion between the two types of relative clauses in 84 Marathi..., and present a transformational analysis of restrictive relative clauses." However, after following her arguments through, one finds that the author has left several things to be desired. For example, Junghare says the follow- ing regarding the formal distinction between restrictive and non-restrictive relative clauses, "... I claim that relative clauses which contain the definite determiner pp or its variants, which mark the scope of antecedents, are restrictive relative clauses. Non-restricted [i.e. non- restrictive] relative clauses, On the other hand, are marked by the absence of pp or its variants ...." (page 253). However, at least in my dialect9 sentences given in (3.4.2) b. (3.4.6) a,(3.4.7) and (3.4.8) in this chapter, reproduced below, give non-restrictive readings with the definite determiner pp : (3.4.2) b. maz; .bhEu jo mumbait rzhgto (to) my brother REL Bombay-in lives that one/he wakil ahe lawyer is 'My brother, who lives in Bombay, is a lawyer.' (3.4.6) a. XSn jo Sikagot rahato (to) John REL Chicago-in lives that one/he 53 ithe yena-r -a'he today here coming is 'John, who lives in Chicago, is coming here today.‘ 85 (3.4.7) jgznni ggthai racali te tukaram maharaj REL-INSTRH Gatha composed thatH St. Tukaram dehula janmale Dehu-at was bornH 'St. Tukaram, who composed the thha, was born in Dehu.‘ (3.4.8) gandhi-nni jglnna guru manale Gandhi-INSTRH REL-toH teacher regardedH te gokhale mawal hote thatH Gokhale moderate wasH 'Gokhale, whom Gandhi regarded as (his) teacher, was a moderate.‘ As said earlier in this chapter, it can be argued that in Marathi the non-restrictive relative clause is not always formally distinct from the restrictive relative clause. Sometimes the former is made distinct from the latter as follows: in the non-restrictive the head NP precedes the relative clause and it is not preceded by the definite determiner. Furthermore, Junghare assumes the Conjunction ana- lysis for non-restrictive relative clauses and the NP-S analysis for restrictive relative clauses in Marathi. In her study Junghare makes certain assumptions without providing any real justification. For example, following Ross (1970) she assumes Marathi to be an SVO language.10 She also assumes the following questionable set of base rules for Marathi: 86 (i) PP-v P NP (ii) NP-h- NP s AS mentioned earlier in this chapter, Marathi is a post- positional language, yet Junghare assumes for it the base rule given in (i) which is applicable to English, which is a prepositional language. Furthermore, she assumes the base rule given in (ii) "in order to handle Marathi relative and complement constructions...." (p. 254), even though in the case of both of these constructions the dominant surface order is subordinate clause followed by main clause (see complement constructions given in (3.1.24) a and b, given in this chapter). In particular, by assuming the underlying NP-S order for restrictive relative clauses in Marathi, Junghare is forced to write certain transformations which are not independently motivated. This point will become clear in the following discussion. According to Junghare (pp. 255-257) in Marathi the following transformational rules in their given order are required to derive restrictive relative clauses from their underlying NP-S structure: 87 Antecedent-Copyipg: NP [ X NP Y I Z S S 1 2 3 4 5 obliga- tory 1 2 3 4 # 1 5 Condition: 1=3 Wh- Attachment: This rule attaches the relative pronoun j; to the NP which is dominated by an embedded S. Noun-Deletion:11 [Lg (Adj)n (n) [x j- t__ (Adj)n (n) Y]] NP 3 S NP 1 2 3 4 5 6 7 8 dopt i onal 1 49 49 4 5 6 7 8 Condition: 2:6, 3=7 Pronominalization: A. Pronominalization from left to right: [x NP Y S[ NPJS z] NP NP 1 2 3 4 5 obliga- =". tory 1 2 3 4 5 [+ Pro] Condition: 2=4 88 B. Pronominalization down in subordinate clauses: x [[NP] Y NP] 2 S S S S 1 2 3 4 5 obliga- "dbtory 1 2 3 4 5 [+ Pro] Condition: 2:4 Extraposition transformation: this rule extraposes the embedded S to the end of the main S. to- Deletion: x to [Y jo z] w s s 1 2 3 4 5 6 z-Dopt i onal Verb-Final: X V Y 1 2 dobligatory 1 ‘T 3+2 The verb is moved to the end of its own S. 19221312122: The relative pronoun jg is Optionally deleted. Note that Of the above transformations, the Antecedent- Copying transformation and the pp deletion transformation are required solely because Junghare assumes the NP-S order. 89 Thus given. NP --P NP 3 the Antecedent-Copying transformation produces the sequence, 1P S NP. The N(P) deletion and £9 deletion transformations Yield _S___N_P. Thus essentially what the above three trans- formations (i.e. Antecedent-Copying transformation, N(P) deletion transformation and _tp deletion transformation) do is reverse the order from NP-S to S-NP. Similarly, the sole Purpose of the Verb Final transformation is to move the verb to the position where it occurs in a sentence. Furthermore, in Junghare's analysis some instances of transformationally derived _t_9_ occur as a result of the N(P) deletion transformation, whereas some other instances of 1:2 occur as a result of the Pronominalization transformation. To me this distinction is spurious, to say the least. (As mentioned earlier in this chapter, it is assumed in the pre- Sent study that all the instances of the transformationally d . o e o erlved pronoun 1n Marathi ar1se from a common source. '1: . hat 18, from ANAPHORIC DEFINITIZATION. TRANSFER OF NOUN FEATURES, and NP DELETION.) The other study on relative clauses in Marathi I :viah to mention is Kelkar (1973). This is presented as '6.“ alternative analysis" (p. 274) to Junghare (1973). The author presents substantial data covering adjectival and adverbial relatives, multiple headed relative clauses, ‘DII-restr1ct1ve relative clauses, etc. ‘ 90 This study may be viewed as quasi-transformational, in that Kelkar uses the transformational format (e.g. tree diagrams and transformational rules), but he is primarily interested in the description of surface structure facts: no attempt is made to look for’ an explanation Of any of these facts. This is understandable for the author is a structuralist in training and persuasion, which is clearly reflected in his remarks (page 298), "First, language universals must be hardwon to be genuine. Methodo- logically, we must proceed as if we are not very hopeful or even anxious to find language universals... A related point is the recent neglect of 'surface'-- possibly a pernicious result of taking the deep-surface metaphor too seriously." In summary, section 1 describes some typological characteristics of Marathi. Sections 2 and 3 present an analysis of restrictive relative clauses in Marathi. It is proposed that the presence of the remote demonstrative in a relative clause serves as the surface structure clue that enables the hearer to relate the head NP to the preceding relative clause. Section 4 presents an analysis of non-res- trictive relative clauses in Marathi. It is shown in this section that in Marathi,non-restrictive relative clauses are not always formally distinct from restrictive relative clauses. In section 5 there is a brief discussion of two previous studies on relative clauses in Marathi, and it is shown that these studies are not adequately explanatory. 1. GMHER3 FOOTNOTES I use these terms, following Andrews (1972). Apparently, there are counter examples to this constraint For consider the following example (I thank Dr. Yamuna Kachru for providing me with this example.). jo he kam karu Sakel asa' ek manus mala REL this work do can such that one man to me mahit ahe known is 'I know a man who can do this job.‘ In this sentence with the correlative pronoun as; 'such that' in it, the referent for the head NP is presupposed to exist. This is due to the presence of the factive verb 'know' in the sentence. I suspect that this temporal order of background vs. new information is significant in the Speech perception process. In other words, one of the perceptual strategies employed by the hearer might be to look for or to anti- cipate the background information temporally prior to the new information. That explains, I think, why extra- posed relative clauses (Ziv and Cole. 1974) are inter- preted as containing the new information (i.e. the 91 92 assertion). Chafe (1974) makes similar observations in this regard. The following examples are from Ziv and Cole (1974. p- 773). a. A man who had three ears came into the room. b. A man came into the room who had three ears. In sentence (a) the main clause is perceived to be introducing the new information (i.e. the assertion): whereas in sentence (b) the extraposed relative clause is perceived to be introducing the new information (i.e. the assertion). This condition is required to block such ungrammatical sentences as the following: *to manus lekhak ahe kal 515 COREL man writer is yesterday came M. Smith (1974) proposes a continuum of variable reference, somewhat similar to this scale. Sentences (3.4.5) b and c are odd since we normally expect a person to have one wife, one father etc. However, among certain communities in North India, a woman may have more than one husband (and these are brothers). In fact, in the Indian epic, the Mahgbharata, the five brothers (i.e. the Pandawas) have a single wife, Draupadi. Loetscher (1973) makes Similar Observations in this regard. 8. 10. 11. 93 I do not know of any other studies in existence in the linguistic literature that deal with relative clauses in Marathi within the framework of a modern linguistic theory. I have consulted other Speakers of my dialect and they also get the non-restrictive readings for these sentences. Ross (1970) claims that there is a correlation between the directionality of gapping and the basic word order of a language. However, since the publication Of this work, several studies have appeared (e.g. Maling, 1972). which conclusively disprove Ross' arguments. In fact. Ross himself (see Ross, 1973) retracts his earlier claims and makes the following new claim,"My tentative - I hope it has been Obvious throughout that all of the above is very preliminary - conclusions are that all languages are underlyingly verb-final, but that this deep order can be "loosened" to different degrees in different languages . . . ." (p. 415) It would be more apprOpriate to call this NP deletion since it deletes ggj as well as N. Also, Junghare does not explain how 39 (1 Of SD) suddenly surfaces in this rule. It cannot be the case that the antecedent of every relative clause is definite, for in that case the restric— tive relative clauses with indefinite head NPS cannot be accounted for. CHAPTER 4 RELATIVE PARTICIPLE PHRASES IN MARATHI 4.0 This chapter is divided into the following sections: 4.1. Formation of relative participle phrases in Marathi 4.2. Relative participle phrases and perceptual complexity 4.1 Formation of relative participle phrases in Marathi As mentioned in chapter 3 there is a proliferation of so-called relative participle phrase constructions in Marathi. It was also noted there that this could be due to the Dravi- dian influence. This does not mean, however, that this type i of construction has come into Marathi solely from a Dravidian source, Since Sanskrit also has this type of construction, but it is not as frequent there as it is in Marathi. In essence, these relative participle phrases are verbal adjec- tives. For example, consider the following sentences in English: (4.1.1):a.The man who is reading the newspaper is my uncle. ‘b.The man reading the newspaper is my uncle. Note that in sentence (4.1.1) a the underlined part is a relative clause, while in sentence (4.1.1) b the underlined 94 95 part may be called a relative phrase, which modifies the noun 'man'. Sentence (4.1.1) b is derived from sentence (4.1.1) a by applying;RELATIVE CLAUSE REDUCTION. Next consider the following sentences: (4.1.2) a. The man who is tall won the race. b. *The man tall won the race. c. The tall man won the race. Sentence (4.1.2) b is derived from sentence (4.1.2) a by applying RELATIVE CLAUSE REDUCTION. Notice that in sentence (4.1.2) b the underlined residual part of the relative clause must be moved in front of the head noun mgp, while in sentence (4.1.1) b, the residue of the relative clause must not be moved in front of its head noun. In English, this movement rule, which may be called MODIFIER SHIFT,must apply if the residual part of the relative clause is a single adjectivel. Now consider the following sentences in Marathi: (4.1.3) a. jo manus unca ahe tya-n1 ti REL man tall is he-INSTR that V saryat jinkali race won 'The man who is tall won the race.‘ - - - v b. tya unca manasa-ni ti saryat jinkali that tall man-INSTR that race won 'The tall man won the race.’ 96 (4.1.4) a. jo minus wartmin patra wicat ihe to REL man newspaper reading is that minus mizi. kiki ihe man my uncle is 'The man who is reading the newspaper is my uncle.‘ b. to wartmin patra wicat asniri} aslela that newspaper readingi BE-PARTICIPLE - u- -- - manus maza kaka ahe man my uncle is 'The man who is reading the neWSpaper is my uncle.‘ If we compare the underlined relative clauses in sentence (4.1.4) a and the underlined relative participle phrase in sentence (4.1.4) b, we see that in the latter sentence, the relative pronoun and the relativized NP are deleted. Note that in contrast to English. in Marathi since the relative clause precedes the head NP, MODIFIER SHIFT is not required and that the single modifier ppgg 'tall' in (4.1.3) b as well as the modifier phrase wartmin patra wicat{asniri aslela 'reading the newspaper' in (4.1.4) b precede their head nouns. This modifier phrase may be called the relative participle phrase and sentences with such phrases may be called relative participle phrase constructions. Furthermore, in the former the verb is in the finite form, while in the latter it is in the non-finite form. The non-finite form of a verb is called the participle form. That these participles 97 are in fact adjectives is clear from the following: (i) (ii) E.g., (4.1.5) palaniri run-PART 'The horse that runs' palaniri run-PART they describe some property of the head noun they agree in number and gender with that noun. gho‘éa' horse ghOPi horse fem 'The horse that runs' - hv palanare g ore run-PART horses 'The horses that run' palaniryi gh run-PART v- orya horses fem 'The horses that run' In Marathi one verbal stem may have several participle forms, e.g. (4.1.6) pala-ne 'to run' — a. h VI. a. palanara g ora run-PART horse 'the horse that runs' b. palat asleli h V- . . {asniri} g ora 'the horse that 18 . running running BE-PART horse c. palileli ghoii run-PART horse 'the horse that ran' a- - h v. d. palanar aslela g ora ' 'the horse that run-FUTURE BE-PART horse will run' The negative forms Of these participles are formed by adding the negative particle pg before them, e.g. 98 (4.1.7) a. na palaniri ghogi' 'the horse that does not I not run-PART horse run b. na palat asleli h V- ' {asniri} g ora not running BE-PART horse or u v- palat naslela_ ghora 'the horse that is ‘ nasanara not running' running not BE-PART horse 0. na palileli ghOii' 'the horse that did not not run-PART horse run' a - hy— d. na .palanar aslela g ora not run-FUTURE BE-PART horse or - - V- palanar naslela ghora 'the horse that run-PART not-BE-PART horse W111 "Ot run' In Marathi there are two types of participles: (1) Active participles: The internal structure of the active participles is as follows: (4.1.8) Active participle *{Verb stem} + PARTICIPLE Aux stem” MARKER + CONCORD MARKER e.g. (4.1.9) a. p a l.a - n a r - a ghogi. 0 o I run horse the horse that runs' Lverb stem- PART M -CONCORD M Participle 99 b. palat a s - l e l - i ghoii 'the horse that running BE horse 18 running' Aux - PART M-CONCORD M ,_stem Participle (ii) Passive participles: In order to understand the internal structure of the passive participles in Marathi. it is necessary to see how the passive construction is derived in Marathi. Consider the following sentences: (4.1.10) a. ibn te pustak wicato John that book reads 'John reads the book.‘ b. Kan-kaiun te pustak wicala jita Johneby that book read goes past participle 'The book is read by John.' As shown in sentence (4.1.10) b above, in the passive construc- tion in Marathi. (i) the underlying subject takes the postposition keratin 'by' (ii) the main verb is changed to its past participle form. (iii) the form of the verb jinp 'to go' is placed after the main verb. (iv) the main verb as well as the form of jing 'to go' agree in number and gender with the under- lying object. Notice that unlike in English, in 100 Marathi the subject and Object do not switch positions. The internal structure of the passive participle: (4.1.11) a. Verb + perfective + CONCORD :5 + PART + CONCORD stem suffix M go M M (4.1.11) b. jbnkagun wicala jinira pustak John-by read going book 'The book read by John.' Next, consider the following sentences: (4.1.12) a. kil jo minus ili to minus lekhak ihe yester- REL man came that man writer is day b . 1:211 j o Q 312 " .. .. .. C . kil Q Q 51;: " " " n d . kal Q § 5.191; n n u I: Sentences (4.1.12) a through c are relative clause construc- tions, while sentence (4.1.12) d is a relative participle phrase construction. It is clear from the above sentences that the latter is derived from the former by means of the deletion of the relative pronoun as well as the relativized NP, and by changing the verb, from its finite form to the non-finite (i.e. the participle) form. In other words, rela- tive clause constructions as well as relative participle phrase constructions are derived from a common underlying structure by means of the following transformational rules: 101 ANAPHORIC DEFINITIZATION RCF TRANSFER OF NOUN FEATURES NP DELETION RELATIVE PRONOUN DELETION LEFT DISLOCATION EXTRAPOSITION RIGHT DISLOCATION RELATIVE PARTICIPLE PHRASE FORMATION The RELATIVE PARTICIPLE PHRASE FORMATION (RPP) rule in Marathi may be formulated as follows: (4.1.13) I x v ] t- NP y S S NP NP SD 1 2 3 4 5 optional :— SC 1 2 4 5 3 [+ PARTICIPLE] Condition: 1 contains neither a relative pronoun nor an NP coreferential with the NP in h. Note that the condition specified for REP above is required since, as shown in sentence (h.1.12) d, which is the relative participle phrase construction, the relative pronoun as well as the relativized NP are deleted. Thus the condition requires that both RELATIVE PRONOUN DELETION and NP DELETION must precedeVRPP. If they were to fellow it, both of these rules, which are Optional, would have to be made more complicated in that they would have to be 102 allowed to 'look back' in the derivation so that they would be obligatorily applied if RPP had preceded them. Otherwise the following ungrammatical sentences would be generated: (4.1.14) a. * kal jo manus i alela to manusi yesterday REL man come-PART that man lekhak ahe writer is b. *jo manus i wartmin-patra wacat asnar; REL man newspaper reading BE-PART manusi maza kaka ahe 0 man my uncle is Note that RPP does not have to be crucially ordered with respect to LEFT DISLOCATION, EXTRAPOSITION and RIGHT DISLOCATION. For example, RPP and LEFT DISLOCATION, in whichever order they are applied, yield the correct output: (4.1.15) given, jo manus kal 515 REL man yesterday came to minus lekhak ahe that man writer is NP DELETION RELATIVE PRONOUN DELETION LEFT DISLOCATION --> to m-a-rlus 1:51 'a'lela that man yesterday come- PART to lekhak ahe that writer is In the above sentence, even though the participle 103 phrase R21 alel; follows to minus, it precedes the deter- miner (and the deleted head NP ). Thus the modifier + N order is still preserved. Similarly RPP may either precede or follow EXTRA- POSITION and RIGHT DISLOCATION. and yet may produce the correct output: (4.1.16) given, jo manus kal £15 REL man yesterday came to manus lekhak ahe that man writer is NP DELETION RELATIVE PRONOUN DELETION EXTRAPOSITION and RIGHT DISLOCATION ":- to maqus lekhak -ahe that man writer is 1:31 519.12 to yesterday come-PART that As in English, in Marathi prenominal adjectives may be derived from relative clauses. However, note that unlike English, the adjectivalization process in Marathi goes through an intermediate stage which is perfectly grammatical. For example. consider the following sentences: . u h - . V "’ (4.1.17) a. JO manus latt a hota to t1 saryat harala REL man fat was he that race lost 'The man who was fat lost the race.‘ 104 b. to lattha aslela minus ti saryat harala that fat BE-PART man that race lost 'The fat man lost the race.‘ c. to lattha minus ti saryat harala the fat man that race lost 'The fat man lost the race.’ The application of RPP to sentence (4.1.17) a produces sen- tence (4.1.17) b: sentence (4.1.17) c is derived by applying PARTICIPLE DELETION, which may be formulated as follows: F “'1 (4.1.18) X +VB +VB t- NP -V +V V +PART V L. NP " NP SD 1 2 3 4 5 option- SC 1 2 (Q a 5 The above rule converts the structure h ‘ - h - latt a aslela to manus _g. latt a to manus _fat BE-PART that man fat that man Since in Marathi, the determiner in the NP must always precede the descriptive adjective (see (3.1.22) in chapter 3), a late level rule yields the correct order of the adjective lattha and the determiner 39. To sum up, it is prOposed above that in Marathi. relative clause constructions and the corresponding relative participle phrase constructions be derived from the same 105 underlying structure. Furthermore, it has been shown that they require the same transformations except that in the case of the relative participle phrase construction. NP DELETION and RELATIVE PRONOUN DELETION are obligatory, while in the case of relative clause constructions they are Option- al. Also, prenominal adjectives are derived from relative clauses through relative participle phrases. Let us turn now to a discussion of some recent work by Keenan and Comrie (1972) on relative clauses. According to them, (p. 1) "languages differ with regard to the possibility of forming restrictive relative clauses on basic NP positions in unmarked simplex.sentences." They give the following hierarchy of basic NP positions: (4.1.19) Accessibility Hierarchy (AH)2 Subj > D obj > I obj 7 Obl 7 Gen 7 0 Comp (where 'Subj'=subject, 'D obj'=direct object, 'I obj'=indirect object, '0bl'=other oblique case NPS, 'Gen'=adnominal genitives, and'O Comp'= object of comparison, ':"="higher on AH than".) Given the above AH, Keenan and Comrie (1972, p.9) claim, "For unmarked simplex sentences, [(4.1.20) a] any basic RCF strategy that can relativize some NP positions on the AH can relativize all NP positions higher than it on the AH and b. for each NP position on the AH there are lan- guages having a basic RCF strategy that can relativize it but cannot relativize any positions lower than it on the AH." 106 In Marathi all of the basic NP positions may be relativized, e.g.. (4.1.21) a. (Subj. NP): be (D Obj NP): c. (I obj NP): d. (Obl. NP): e. (Gen. NP): f. (0 Comp): jo kal '21; to manus REL yesterday came that man 'The man who came yesterday.‘ jon-ni {115' pahila ti mulagi John-INSTR REL-to saw that girl 'The girl who John saw.’ jam-ni 3513 pustak dila John-INSTR REL-to book gave to mulaga that boy 'The boy whom John gave the book to.‘ jbn-ni jalni Zmba John-INSTR REL-with mango kapala ti suri cut that knife 'The knife which John cut the mango with.‘ XECa pustak harawala REL-of book lost to manus that man 'The man whose book was lost.’ jigapeksa meri unca ahe ti mulagi REL-than Mary tall is that girl 'The girl who Mary is taller than.‘ 107 As illustrated above, in Marathi all the basic NP positions on the AH ( Keenan and Comrie, 1972) may be relativized. However, the same cannot be said of the relative participle phrase formation process. In other words, RELATIVE PARTICIPLE PHRASE FORMATION in Marathi is subject to some constraint(s). In order to see what these constraints are, let us consider the following examples. First we will consider the cases in which relative clause formation (RCF) as well as relative phrase participle formation (RPP) are possible. (4.1.22) a. (RCF) jo at gel; to miqus REL in went that man 'The man who went in' b. (RPP) to at gelela minus that in go-PART man 'The man who went in' (4.1.23) a. (RCF) ja-ni joneli marala to manus REL-INSTR John-to beat that man 'The man who beat John' b. (RPP) to 35n-1a mara1e1a Rages that John-to beat-PART man 'The man who beat John' In the above examples the head NP is the subject of the relative clause. 108 (4.1.24) a. (RCF) an-ni 3313 thila to minus John-INSTR REL-to saw that man 'The man whom John saw' b. (RPP) to jBn-ni pahilela manus that John-INSTR see-PART man 'The man whom John saw' (4.1.25) a. (RCF) yon-hi jani patra lihila John-INSTR REL-with letter wrote te pen that pen 'The pen with which John wrote the letter' b. (RPP) te an-ni patra lihilela pen that John-INSTR letter write-PART pen 'The pen with which John wrote the letter' The above examples show that the relative clause may be reduced to the corresponding relative participle phrase if the subject NP of that clause is either coreferential with the head NP or has an overt case ending. The relative clauses which do not meet either of these conditions cannot be reduced to the corresponding relative participles, e.g. (4.1.26) a. (RCF) jsn 3515 pahato to. minus John REL-to sees that man 'The man whom John sees' b. (RPP) *to XSn pahanara manus that John see-PART man 'The man whom John sees' 109 In (4.1.26) a the subject NP, jen 'John' is neither corefer- ential with the head NP, to minus 'that man' nor does it have an overt case ending. Therefore, the relative clause in (4.1.26) a cannot be reduced to the corresponding relative participle phrase, as shown in (4.1.26) b. In order to see why (4.1.26) b is ungrammatical, let us compare it with (4.1.24) b, reproduced below. (4.1.24) b. to an-ni pahilela manus that John-INSTR see-PART man 'The man whom John saw' In (4.1.24) b, the ending :3; identifies the NP 'John' as the underlying subject of the relative clause, whereas in (4.1.26) b there is no overt marker on the NP 'John' identi- fying it as the subject NP. As mentioned earlier, the relative participle phrase formation process involves the deletion of the relativized NP and the relative pronoun. In (4.1.24) a, reproduced below, (4.1.24) a. HSn-ni 3515 pahilg to minus John-INSTR REL-to saw that man 'The man whom John saw' the relative pronoun 35;; 'whom' is marked for the case which denotes the underlying function of the relativized NP. Notice that even though this relative pronoun is not present in (4.1.24) b, the ending :3; on the NT’Eén provides a surface structure clue which identifies iég as the underlying subject, and the following reading is obtained: 110 'The man whom John saw' 0n the other hand, in (4.1.26) b, reproduced below, (4.1.26) b. *to gen pahanara minus that John see-PART man 'The man whom John sees' there is no surface structure clue to identify the underlying subject in this construction. In order to derive a grammati- cally correct relative participle phrase from (4.1.26) a, the PASSIVE rule must be applied prior to the formation of the participle phrase: (4.1.26) a. 35m 3515' pahato to manus John REL-to sees that man 'The man whom John sees' (4.1.27) a. PASSIVE III-I-dh gen-karun 3515 pahila jata John-by REL-to seen goes to manus that man 'The man who is seen by John' b. RPP V ’ — - O - .- ===-=ID to jan-karun pahila janara manus that John-by seen go-PART man 'The man who is seen by John' The PASSIVE rule adds the ending karun 'by' to the subject NP Sin 'John' of (4.1.27) a. This satisfies the condition for the formation of relative participle phrases: the subject NP of a relative clause must have an overt case ending. Notice 111 that the ending gafign 'by' in (4.1.27) a marks the NP 323 'John' as the underlying subject of that construction. The presence of this ending allows the deletion of the relative pronoun fiélé 'whom', hence the relative participle phrase in (4.1.27) b may be generated. The additional examples given below further illustrate the point just made above, namely that in the formation of relative participle phrases the functional information must be preserved and may on occasion include in addition to subject and object relations, the other specific concepts expressed by postpositions such as 'without','along with',etc. (4.1.28) a. (RCF) jan {aaagi bhandala to mulag-a John REL—with quarreled that boy 'The boy who John quarreled with' b. (RPP) *to 35n bhandalela mulaga that John quarreled-PART boy 'The boy who John quarreled with' Notice that in sentence (4.1.28) a the relative pronoun is marked for the case which denotes the underlying function of the relativized NP and the subject NP jig 'John' is without any overt case ending. So when the relative pronoun is deleted as shown in sentence (4.1.28) b, it is difficult to perceive which of the two NPs (51g 'John' and mulagi 'boy') is the subject and which one is the object. Hence in order to derive the grammatically correct relative participle phrase from sentence (4.1.28) a, the PASSIVE rule must precede RPP. 112 This is insured by the fact that RPP applies at the higher cycle. (4.1.28) a. jun 5353i bhandala to mulagi John REL-with quarreled that boy 'The boy who John quarreled with' (4.1.29) a. PASSIVE ---t155n-karun j§55§i bhandala gela to mulagi John-by REL-with quarreled went that boy 'The boy who was quarreled with by John' (4.1.29) b. RPP —-P to Han-karun bhandala gelela mulagi that John-by quarreled went-PART boy 'The boy who was quarreled with by John' The PASSIVE rule marks the subject NP 35h 'John' with the ending karun 'by', thus allowing the deletion of the relative pronoun jaéisi 'with whom'. Next consider the following examples: (4.1.30) a. Sen-ni ji-ni Emba kapala _ ti suri John—INSTR REL-with mango cut that knife 'The knife John cut the mango with' b . ti 3:: n-ni 'ambE kapaleli suri that John-INSTR mango cut-PART knife 'The knife John cut the mango with' (4.1.31) a. jam-hi jiEELsiwfiy §mb5 kfipala ti suri John-INSTR REL-without mango cut that knife 'The knife without which John cut the mango' (i.e. The knife John did not use to cut the mango with) 113 b.*ti jon-ni Emba kapaleli suri that John-INSTR mango cut-PART knife 'The knife John cut the mango with' In (4.1.30) b, the relative pronoun with its postposition (i.e.‘Xi;gi_'with which') is deleted, and yet the meaning is not changed, that is, (4.1.30) a and b have the same meaning. 0n the other hand, if the relative pronoun with its postposi- tion is deleted from (4.1.31) a, the meaning is changed as seen in (4.1.31) b. This means that the deletion of the rela- tive pronoun with its postposition is not allowed in (4.1.31) a, thus blocking the derivation of the relative participle phrase. There is a possible explanation as to why the dele- tion in this particular case is not allowed: the deletion of jigasiwiy 'without which' leads to a non-recoverable deletion in (4.1.31) b and therefore changes the meaning of the parti- ciple phrase.3 The same is not the case with (4.1.30) a from which 1i1gi 'with which' is deleted to derive the participle phrase, given in (4.1.30) b. Notice that (4.1.30) b and (4.1.31) b would have identical surface structures, even though they would have come from different structures with different meanings (i.e. (4.1.30) b would come from (4.1.30) a and (4.1.31) b would come from (4.1.31) a). Another example in which the deletion of the relative pronoun with its postposition is disallowed is as follows: (4.1.32) a. an-ni jizabarobar pustak pathawala ti mulagi John-INSTR REL-with book sent that girl 'The girl with whom John sent the book' 114 b.*ti jbn-ni pustak pathawaleli mulagi that John-INSTR book send-PART girl 'The girl to whom John sent the book' Thus the above examples show that the deletion of the rela— tive pronoun with its postposition, EEEE-barobar 'with whom' changes the meaning of the sentence. It appears from the examples given in (4.1.30) through (4.1.32) that REL PRONOUN DELETION must be made 'subject to some constraint that will allow the deletion of relatiye pronouns with certain'postpositions only (.i.e.,with unmarked cases, such as the agentive and objective).. To sum up, therefore, the formation of relative participle phrases in Marathi is subject to the following condition: (4.1-33) The formation of relative participle phrases is possible in Marathi if (and only if) the subject NP of a relative clause (i) is coreferential with the head NP or (ii) has an overt case ending. For example, the subject NP of the relative clause in (4.1.22) a is coreferential with the head NP, hence it is possible to derive the corresponding relative participle phrase, given in (4.1.22) b. Similarly, since the subject NP of the relative clause in (4.1.23) a has an overt case ending, 115 the corresponding relative participle phrase, given in (4.1.23) b is possible. In contrast to this, the subject NP of the relative clause in (4.1.26) a is neither coreferential with the head NP nor does it have an overt case ending, hence the resulting relative participle phrase would be ungramma- tical, as shown in (4.1.26) b. In order to derive the grammatically correct relative participle phrase from (4.1.26) a, therefore, PASSIVE must be applied prior to the application of RELATIVE PARTICIPLE PHRASE FORMATION, as shown in (4.1.2?) a and b. The PASSIVE rule adds an overt case ending to the subject NP of the relative clause, thus satisfying the condition for the forma- tion of relative participle phrases. Notice that it is not possible to effect a change transformationally in order to alter a structure to meet condition (i) given in (4.1.33), because such a change would affect the semantic representation itself. However, condition (ii) is fairly superficial in nature, hence it is possible to effect a change transforma- tionally so as to alter a structure to meet this condition. The functional explanation for condition (4.1.33) lies in the fact that the deleted information must be recoverable from the surface structure of a sentence for its correct semantic interpretation. 116 4.2 Relative participle phrases and perceptual complexity Chomsky (1965) suggests that there are certain aspects of the structure of a sentence that affect the per- ception of that sentence: (1) repeated nesting contributes to unacceptability (ii) self-embedding contributes still more radically to unacceptability (iii) multiple branching constructions are optimal in acceptability (iv) nesting of a long and complex element reduces acceptability (v) there are no clear examples of unacceptability involving only left-branching or only right branching (page 13) In this section it is shown that in Marathi the. relative participle phrase construction plays a significant role in reducing the perceptual complexity of certain struc- tural configurations.“ In Marathi, adjectives, complement and relative clauses precede the head nouns. This characteristic of Marathi makes it a left-branching language. For example, consider the following sentence. (4.2.1) jfin-ni XELIE pahila tya mula-ni John-INSTR REL-to saw that boy-INSTR 35-13 marala tya kutrya-ni 3505 pithalag REL-to hit that dog-INSTR REL-Of chase 117 kal; te manjar palzla did that cat ran 'The cat that the dog that the boy that John saw hit chased ran.‘ Notice that sentence (4.2.1) is perfectly grammatical and can be very easily understood by the Marathi Speaker, while_ its English translation is almost unintelligible in that the syntactic structure of the English sentence involves self- embedding. The surface structure of sentence (4.2.1) may be represented as follows: (4.2.2) S 0 A NP1. VP /\ V. 31 NF} palala te manjar ran- that cat aca path alag kela REL-of chase did A tya kutm ani that dog-INSTR A jala marala REL-to hit NP tya mulani that boy-INSTR S ibn-ni 3315 pahila John-INSTR REL-to saw 118 Notice that in the above structure three sentences are embedded into the main sentence, and yet the entire sentence can be interpreted without any difficulty whatsoever. This is due to the fact that the above structure is an instance of left branching. Next consider the following sentence: (4.2.3) jbn-ni tya manasa-Ia jaLni to John—INSTR that man-to REL-INSTR that lanqaga jalni tya kutryaca pgthalag fox REL-INSTR that dog-of chase kela‘ to marala tyalg pahila did that one killed that one-to saw 'John saw the man who killed the fox that chased the dog.‘ The syntactic structure of the English translation of the above sentence does not pose any problem in English: (4.2.4) So NP VP John ,z”fl“‘\\‘ Y NP saw ’///’~\\\’ NP 81 NF VP whol /\ Y NP killed /\ S NP .2 AA. the fox. that.chased the J Jdog 119 This is a right branching construction, which is one of the characteristics of English, and therefore the sentence does not involve any perceptual complexity. However, the syntactic structure of the Marathi sentence given in (4.2.3) invloves self-embedding, as shown in the following structure: (4.2.5) So A NP VP A Hen-n1 John-INSTR NP Y pahila saw NP 31 tya m'anasa-l'ai the man-to ’ NP VP v4 . janii REL-INSTR /\ NP Y marala killed NP S2 to landagaj jalni. tya kutryaca the fox REL-INSTR the dog-of pathalag kela chase did The self-embedding nature of the above structure makes the sentence perceptually quite complex for the Marathi speaker. However, the relative participle phrase construction derived from sentence (4.2.3) reduces its structural complexity and thus makes it perceptually simpler: 120 (4.2.6) an-ni kutryacg pathalag kelel; landaga John-INSTR dog-of chase do-PART fox maralelya manasa-la pahila kill-PART man-to saw 'John saw the man who killed the fox that chased the dog.‘ This point will become clear if we compare the structural configuration of sentence (4.2.3), which is a relative clause construction and that of sentence (4.2.6), which is a relative participle phrase construction. (4.2.7) a. an-ni tya manasa-lai jalni to John-INSTR the man-to REL-INSTR the S S 0 1 landagaj jalni tya kutryaca pathalag kela fox REL-INSTR the dog-of chase did 82 82 to marala tya-15:1 p'a'hil-a' that one killed that one-to saw 51 8O W- b. ' jan-ni kutryaca pgthalag kelela John-INSTR dog-of chase do-PART NP NP NP ‘ VP 1.. SO 1 W landaga maralelya mEnasZ-la pahila fox kill-PART man-to saw NP V 'VP - s 121 As shown in (4.2.7) b, the relative participle phrase con- struction reduces the self-embedding structure in (4.2.7) a to a flat structure (i.e. without any embedding), thus making the sentence perceptually much simpler. More examples follow: Consider the following sentences: (4.2.8) SI. tya m§njar§-ni te dudh sandala the cat-INSTR the milk spilt 'The cat spilt the milk.’ 82. tyE undara-ni te dudh anala the rat-INSTR the milk brought 'The rat brought the milk.‘ 83. tya kutryE-ni tya undara-la pahila the dog-INSTR the rat-to 'The dog saw the rat.‘ Embedding 82 and S3 into Si we get, (4.2.9) a. ty; manjara-ni tya undarE-ni the cat-INSTR the rat-INSTR SI 82 kutrya-ni pahil 3e 'Enala dog-INSTR saw S3 ‘82 saw Kat-la- tya REL-to the S3 te dudh'sandala REL brought that milk spilt 'The cat spilt the milk that the rat that the dog saw brought.’ Note that the syntactic structure of the above sentence in Marathi contains self-embedding, which makes the sentence 122 almost unintelligible. However, the corresponding participle construction converts the self-embedding structure into a single sentence: b. L. LS1 tya m§njar5-n' tyE'kutryalni pahilelya the dog-INSTR see-PART the Cat-INSTR NP NP 4p tya undarE-ni 'analela te dudh the rat-INSTR bring-PART the milk Notice that if the three sentences (i.e. 81, NP [sandala [spilt 1] 82, and S3) are embedded into one another as shown in the following structures, the result is an ungrammatical sentence in English, but not in Marathi. -s1 (4.2.10) a. English SI NP VP A the cat /\ V NP I spilt//f\\\\ NP S3 the mir://\\\\ NP VP the 30g ///\\\ Y NP saw NP the rat 123 (4.2.11) a. Marathi 31 ”,/*~““ N2 VP tya mEnjara-ni the cat-INSTR NP V sandala /\ spilt S3 NP 6h te dud the milk NP vp m. tya kutrya-ni the dog-INSTR NP V - l pahila saw NP A tya undara-li the rat-to AA the rat tya undarE-ni te the rat-INSTR the brought the milk dudh 'Enala milk brought b. b. *‘The cat spilt the milk- which the dog saw the rat that brought.‘ tya mEnjarE-ni tya kutryE-ni the cat-INSTR the dog-INSTR jELni je dudh Enala REL-INSTR REL milk brought tya undara-la'pihila te that rat-to saw that dudh sandals? milk spilt 124 The ungrammaticality of sentence (4.2.10) a in English results from the violation of the Complex NP Constraint. In contrast to this, sentence (4.2.11) a in Marathi is perfect- ly grammatical since there is no violation of the Complex NP Constraint in this sentence (For discussion of this and other constraints by Ross (1967), see chapter 5.). In the above examples we have seen that the relative participle phrase construction reduces the perceptual complexity of the structure from which it is derived. However, this is not always true, as the following example shows. Consider sentence (4.2.1) again, reproduced below, along with the relative participle phrase construction (4.2.12) which is derived from it. (4.2.1) saneni 35:15 pahiia tyE mula-ni 35:1; John-INSTR REL-to saw that boy-INSTR REL-to marala tya kutrya-ni Kale; pathalag kela hit the dog-INSTR REL-of chase did te manjar palala the cat ran 'The cat that the dog that the boy thatuJohn saw hit chased ran.‘ (4.2.12) an-ni eahiieiya mulE-ni maraieiya kutryZ-ni John-INSTR see-PART boy-INSTR hit-PART dog-INSTR pathalag kelela manjar payala chase do-PART cat ran 'The cat that the dog that the boy that John saw hit chased ran.‘ 125 Notice that in sentence.(4.2.12) all the embedded clauses from sentence (4.2.1) are reduced to relative participles. Therefore, this sentence has the surface structure configu- ration of a simplex sentence. Nevertheless, contrary to what one would expect, sentence (4.2.12) is g9; perceptually simpler than sentence (4.2.1). The reason for this is that although sentence (4.2.1) is structurally more complex than sentence (4.2.12), the former has only one possible inter- pretation, whereas the latter has more than one possible interpretation, depending on how it is parsed: one of the interpretations is the same as that of sentence (4.2.1). In the other interpretation both of relative participle phrases 'that John saw' and 'that the boy hit' modify the NP 'the dog' since the CONCORD MARKER :a in both phrases agree with that NP. As mentioned earlier in this study, in Marathi some- times a formal distinction is made between restrictive and non-restrictive relative clauses. However, this distinction is never present in relative participle phrases: (4.2.13) a. Restrictive relative clause: jo mumbait rahato to maia bhau wakil ahe REL Bombay-in lives that my brother lawyer is 'My brother who lives in Bombay is a lawyer.‘ b. Non-reStrictive relative clause: miza bhau ' jo mumbait rahato (to) wakil ‘ahe my brother REL Bombay-in lives that lawyer is one 'My brother,who lives in Bombay,is a lawyer.‘ 126 c. Relative participle phrase: to mumbait ran'anara mEz'i than wakil “ahe the Bombay-in live-PART my brother lawyer is either 'My brother who lives in Bombay is a lawyer.' or 'My brother’who lives in Bombay,is a lawyer.’ The ambiguity involved in the above participle construction can be resolved only within the contextual elements of a discourse. To sum up, in this chapter it is proposed that in Marathi,relative clauses and relative participle phrases be derived from the same underlying structure. Furthermore, it is shown that they require the same transformations, except that in the case of relative participle phrases, NP DELETION and RELATIVE PRONOUN DELETION are obligatory, while in the case of relative clauses they are optional. Also, prenominal adjectives are derived from relative clauses through relative participle phrases. Furthermore, it is shown in this chapter that the formation of relative participle phrases is subject to some constraint, which ensures that functional information is preserved in the surface structure of a sentence. In addition, it is shown that relative participle phrases play a significant role in reducing the perceptual complexity of such structural configurations as self-embedding. Lastly, it is pointed out that the formal distinction between restrictive and non- restrictive relative clauses is neutralized in relative participle phrases. CHAPTER 4 FOOTNOTES In Englsh, MODIFIER SHIFT is constrained to prevent an adjective from being moved in front of pronouns, e.g. Something that was strange RELATIVE CLAUSE REDUCTION u—aah Something strange MODIFIER SHIFT '==-'—-'——>* Strange something Based on their study of the formation of restrictive relative clauses in over forty languages, Keenan and Comrie (1972) prOpose the hierarchy of basic NP posi- tions, given in (4.1.19). According to them the higher an NP is on the AH the more easily accessible it is to the relativization process. In sentence (4.1.31) b the deletion of xiaasimay 'without which' is permissible only if the negative form of a participle (e.g. (4.1.7)) is used, e.g. ti Ken-n1 Emba na kapaleli suri that John-INSTR mango not cut-PART knife 'The knife John did not cut the mango with' In recent years, a number of studies have appeared dealing with the perceptual complexity of sentences, e.g. 127 128 Bach (1968), Fodor and Garrett (1967), Fodor, Garrett and Bever (1968, 1975), Kimball (1973), etc. According to Dr. Yamuna Kachru (personal communication), sentence (4.2.11) a is ungrammatical. Admittedly,it is not a simple sentence perceptually. However, I would not label it ungrammatical in my dialect. CHAPTER 5 RELATIVE CLAUSE FORMATION AND QUESTION FORMATION IN MARATHI AND ROSS' CONSTRAINTS 5.0 This Chapter is divided into the following sections: 5.1. RELATIVE CLAUSE FORMATION and QUESTION FORMATION in Marathi 5.2. Relative clauses in Marathi and Ross' (1967) Constraints 5.3. Focus and relative clause constructions in Marathi 5.4. Multiple headed relative clauses in Marathi 5.1 RELATIVE CLAUSE FORMATION and QUESTION FORMATION in Marathi As mentioned earlier in this study, QUESTION FORMATION in Marathi is markedly different from that in English. To recapitulate these differences, in Marathi, yes/no type questions are formed in one of the following two ways: (1) (5.1.1) rising intonation ’4 an-ni te pustak Enala ? John-INSTR that book brought 'Did John bring the book?’ 129 130 (ii) addition of the interrogative pronoun kay at the end of a sentence (5.1.2) XBn-ni te pustak 'Enala kay? John-INSTR that book brought QUEST 'Did John bring the book?‘ The formation of NH questions in Marathi proceeds as follows: the NP to be questioned is replaced by the interro- gative pronoun k;, which agrees with that NP in terms of its case ending. However, in contrast to English, in Marathi the interrogative pronoun is not moved to the front of the sentence, e.g. (5.1.3) a. an-ni til-15 te pustak dila John-INSTR Bill-to that book gave 'John gave the book to Bill.’ b. XSn-ni kunala te pustak dila? John-INSTR whom that book gave 'Who did John give the book to?‘ c. an-ni bil-la kEy dila? John-INSTR Bill-to what gave 'What did John give to Bill?‘ To sum up, in Marathi (i) in both types of questions (i.e. the yes/ho type and WH questions), a WH formative is used (marked by k; in Marathi). (ii) in the case of NH questions, the NP to be ques- tioned is not moved to the front of the sentence. 131 With this background information, let us see how a noun phrase within a relative clause in Marathi is question- ed. Consider the following sentences: (5.1.4) a. jé-ni bil-l; te pustak dila to REL-INSTR Bill-to that book gave that mulaga itha kEm karato boy here work does 'The boy who gave the book to Bill works here.‘ b. ji-ni kunala te pustak dila to REL-INSTR whom that book gave that mulage itha kam karato? boy here work does *‘Who did the boy who gave the book to works here?‘ The surface structure of sentence (5.1.4) a is that shown in (5.1.5) ”'1'” /So\ NP VP A itha kém karato here work does /51\ /”’\ NP VP Art N I - Ha-nii to mulagai the boy REL-INSTR NP NP V - I ' bilLla pustak dila Bill-to book gave 132 In the derivation of sentence (5.1.4) b, the NP bil-la 'Bill-to' is replaced by the interrogative pronoun kunala 'to whom'. As mentioned earlier, QUESTION FORMATION in Marathi does not involve the movement of the NP to be questioned. Notice that the English equivalent of sentence (5.1.4) b is ungrammatical. In Marathi it is possible to form multiple WH questions in the relative clause and to question all the NPs in the relative clause except the relativized one. (5.1.6) ji—ni kunala kay dila to mulagi itha REL-INSTR whom what gave that boy here ram karato? work does *‘Who did the boy who gave what to works here?‘ or *‘What did the boy who gave to whom works here?‘ Sentences (5.1.4) a and (5.1.6) above represent either echo 1 In questions or quiz questions, depending on the context. actual speech, however, these question sentences are more likely to occur without the relative pronoun, or with a relative participle phrase. For example, sentence (5.1.4) b may occur as (5.1.7) a or b. (5.1.7) a. Runala te pustak dila to mulagi itha whom that book gave that boy here kam karato? *‘Who did the boy who gave the book to works here?’ b. 133 to kunala te pustak dilela mulage that whom that book give-PART boy itha kam karato? here work does * 'Who did the boy who gave the book to works here?’ A closer look at question formation within the rela- tive clause in Marathi reveals that this process is subject to some constraint. Consider the following sentence: (5.1.8) to mulagE itha kim karato XEFni that boy here work does REL-INSTR til-15 te pustak dila to Bill-to that book gave that one 'The boy works here, who gave the book to Bill.‘ Notice that in the above sentence the relative clause is moved to the end of the sentence by EXTRAPOSITION. If the NPs in this extraposed relative clause are questioned, the following ungrammatical sentences result: (5.1.9) a. * to mulage itha kam karato! jaLni that boy here work does REL-INSTR Russia te pustak dila (to)? whom that book gave that one * 'Who did the boy who gave the book to works here?‘ * to mulaga itha kam karato 354n1 that boy here work does REL-INSTR til-15 Ray dila to? Bill-to what gave that one * 'What did the boy who gave Bill works here?‘ 134 Notice that the ungrammaticality of the above Marathi sentence does not result from a violation of the Complex NP Constraint, since RELATIVE CLAUSE FORMATION and QUESTION FORMATION are not reordering transformations in Marathi. (See section 2 of this chapter.) While no explanation for the ungrammaticality of these sentences is provided here, I can offer the following speculation: As pointed out by Ziv and Cole (1974), extraposed relative clauses are often inter- preted as containing the new information (i.e. the assertion). In that case it would be infelicitous on the speaker's part to assume that the hearer already knows a part of the new information (and, therefore, would be able to answer a question about it). For note that in sentences (5.1.9) the speaker provides only partial information to the hearer and holds him responsible for the remaining portion of that information. Unlike the 5; questions (i.e. the WH questions), the yes/no type questions in Marathi cannot operate within the domain of the relative clause. One of the differences between these two types of questions is that 5; questions are about noun phrases, whereas yes/no types of questions involve the entire clause. It is often the case that in a relative clause construction the relative clause contains the presupposition of that sentence, e.g. ( 5.1.10) Xa-ni ya kawit'a lihilya to minus wi‘éitra ahe REL-INSTR these poems wrote that man weird is 'The man who wrote these poems is weird.‘ 135 In the above sentence, the relative clause contains a presupposition of that sentence, that is, a proposition that must be true in order for that sentence to be uttered felicitously. In the following sentence, (5.1.11) *jalni ya kawita lihilya kg to minus REL-INSTR these poems wrote QUEST that man 0V. - wlcltra ahe weird is *‘Did the man who wrote these poems is weird?‘ the yes/no type of question is used to question the presuppo- sition of the sentence. This violates felicity conditions since the speaker cannot presuppose something and question it at the same time: hence, the resulting sentence is ungrammatical. 5.2 Relative clauses in Marathi and Ross'(1967) Constraints Ross (1967, p. 65) proposes "a set of constraints, some universal, some language particular...." These con- straints are: The Complex NP Constraint The Sentential Subject Constraint The Coordinate Structure Constraint The Pied Piping Convention According to Ross (1967, p. 65) these constraints "only affect what I will refer to informally as reordering trans- formations-- transformations which have the effect of moving one or more terms of the structural description around some 136 other terms of it." Two examples of reordering transforma- tions given by Ross (1967, p. 65) are QUESTION FORMATION and RELATIVE CLAUSE FORMATION in English. In what follows, these constraints are discussed in order to see if and how they are applicable in Marathi with respect to the rules required for the derivation of relative clauses and questions. The Complex NP Constraint: Ross (1967, p. 70) states this constraint as follows: (5.2.1) No element contained in a sentence dominated by a noun phrase with a lexical head noun may be moved out of that noun phrase by a transformation. Ross represents this constraint diagramatically as follows (Ross 1967, p. 70): (5.2.2) NP /\ N? S [:iiex] N" A M The above constraint accounts for the ungrammaticality of sentences (5.2.3) b (Ross 1967.1» 66) and (5.2.4) b (Ross 1967, P- 69): (5.2.3) a. I read a statement which was about that man. b.*That man who I read statement which was about is sick. 137 (5.2.4) a. Phineas knows a girl who is jealous of Maxine. b.*Who does Phineas know a girl who is jealous of?‘ The surface structure of sentence (5.2.3) a is that shown in (5.2.5). (5.2.5) /SO\ NP VP I Y read NP 81 A Art N n a state- ment NP VP which Y PP was /\ P NP about /\ ' Ari“ 1.) that man In the derivation of sentence (5.2.3) b, the NP 'that man' is moved out of the Complex NP (i.e. the circled NP node in (5.2.5)), thus violating the Complex NP Constraint. Similarly, in the derivation of sentence (5.2.4) b, the Complex NP Constraint is violated. The surface structure of sentence (5.2.4) a is that shown in (5.2.6), 138 (5.2.6) NP’IIEQPPP‘VP Phineas v’//’/\\\\\ knows /NP\SA1 Art a girlNP .¢=::::=. ///£\\\\\ is jealous 1: I)? of Maxine Sentence (5.2.4) b is derived by moving the NP 'Maxine" out of the Complex NP (i.e. the circled NP node in (5.2.6)), hence it is ungrammatical. From the above examples it is clear that in English, it is not possible to either relativize or question the noun phrase from within a relative clause. In contrast to this, in Marathi the noun phrase.within a relative clause may be either relativized or questioned without violating the Complex NP Constraint. The reason for this is that unlike in English, in Marathi RELATIVE CLAUSE FORMATION and QUESTION FORMATION are not reordering transformations. The following discussion will make this point clear. Consider the following examples from Marathi: 139 (5.2.7) a. tya muli-ni jo k 5115' to 'amsa Zmbat hot; the girl-INSTR REL ate the mango sour was 'The mango which the girl ate was sour.‘ h b. Xi-nii joj kh§11§ to ambaj ‘ambat hot; REL-INSTR REL ate the mango sour was ti mulagii Ejari ahe the girl sick is *‘The girl who the mango which ate was sour is sick.’ The surface structure of sentence (5.2.?) a is that shown in (5.2.8). “'2'“ /So\ NP VP A ’////’/h\\\\\\‘ ambat hota sour was S1 NP /\ NP VP Art N ' to amt; the mango Art N tyE muli-ni the girl-INSTR NP V .' h-' — 30 k alla REL ate Notice that in the derivation of sentence (5.2.7) b, the NP tya muli-ni 'the girl' within the relative clause (i.e. SI) is relativized. In English, but not in Marathi, this would constitute a violation of the Complex NP Constraint. 140 This explains why the Marathi sentence given in (5.2.7) b is grammatical, while its English equivalent is ungrammatical. More examples follow. (5.2.9) a. tya'manasa-ni je harawala te pustak khup the man-INSTR REL lost the book very mahag hota expensive was 'The book which the man lost was very expensive.’ b. XELnii Xej harawala te pustakj khup REL-INSTR REL lost the book very mahag hot; to mantisi niskalaji ahe expensive was the man careless is * 'The man who the book which lost was very expensive is careless.‘ The surface structure of sentence (5.2.9) a and b are shown in (5.2.10) a and b,respectively. (502.10) a. SO NP VP A khup mahag hota very expensive was 81 NP / ‘ NP VP Aft N Art//~\\N te pustak that book I _ _ tya manasa-ni the man-INSTR NP V ' I je harawala REL lost 141 (5.2.10) b. S /\ NP VP .-==::::==-. /\ niskalaji ahe careless is 0 NP /\ Art N I - to manus that man NP VP__ {1 khup mahag hota very expensive was S S NP 1 Ant N te pustak that book NP VP NP/\V J I je harawala REL lost Agt N 55 (mana35J-ni REL (man)-INSTR As shown in (5.2.10) b, sentence (5.2.9) b is derived by relativizing the NP tya minasi-ni 'the man' from within the relative clause (i.e. Si): and yet the sentence is grammatical, while its English equivalent is ungrammatical. Sentences (5.2.7) b and (5.2.9) b show that in Marathi it is possible to form a relative clause construction in which the relativized NP is two sentences down the head NP. Notice that these sentences cannot be generated by RCF given in (3.3.2). This rule needs to be revised as follows: 142 (5.2.11) 392 (revised) [ W NPi X J Y NPi S '8 SD 1 2 3 4 5 SC 1 j+2 3 4 5 ’ Condition: 1 does not contain any NP coreferential with 2 of SD. Further, the Complex NP Constraint, given in (5.2.1), does not allow the movement of a noun phrase from within fact-S constructions in English (Stockwell et al. (1973)). Ross (1967, p. 70) gives the following examples. (5.2.12) a. I believed the claim that Otto was wearing this hat. b.*The hat which I believed the claim that Otto was wearing is red. The surface structure of sentence (5.2.12) a is Chat shown in (5.2.13). 143 (5.2.13) S0 A NP VP I Y GD believed /NP\ 81 Agt N the claim NP VP that O'tto /\ V NP 1% /\ wearing Aft N this hat Notice that in the derivation of sentence (5.2.12) b, the NP 'this hat' is moved out of the complement clause (i.e. SB which is a part of the Complex NP (i.e. the circled NP node in (5.2.13)). This results in.a violation of the Complex NP Constraint, hence the sentence is ungrammatical. In contrast to this, in Marathi it is possible to relativize the noun phrase within an embedded clause in the fact-s constructions. This can be seen from the grammati- cality of sentence (5.2.14) b. (5.2.14) a. jhn-lé meri tya mEnasa-barobar geli hi John-to Mary the man-with went this gosta awaPali nahi story liked not 'John did not like the fact that Mary went out with the man.' 144 b. XSn-IE' meri XECE-barobar baher geli hi John-to Mary REL-with out went this gogta EwaPali nfihi to minus grimanta ahe story liked not that man rich is *‘The man who John did not like the fact that Mary went out with is rich.’ The surface structure of sentence (5.2.14) a is that shown in (5.2.15). (5.2.15) SA NP VP jcn-li ’/,//fl\\\ John-to NP V fiwafali nahi liked not S1 I NP A|r(\N hi gosta this story meri tya manasa-barobar Mary the man-with bEher geli out went In the derivation of sentence (5.2.15) b, the NP tyE’mEnasi- bargbar 'man with' within the complement clause is relati- vized. Notice that this sentence is grammatical, while its English equivalent is ungrammatical. In summary, in contrast to English, in Marathi it is possible to either relativize or question an element within a sentence dominated by a noun phrase with a lexical head 145 noun (i.e. the Complex NP). The reason for this is that RELATIVE CLAUSE FORMATION and QUESTION FORMATION are re- ordering transformations in English, but not in Marathi where these processes do not involve movement. The Sentential Subject Constraint: Ross (1967, p. 134) states this constraint as follows: (5.2.16) No element dominated by an S may be moved out of that S if that S node is immediately dominated by an NP which itself is immediately dominated by 8. Ross gives the following examples (1967, p. 135): (5.2.17) a. That I brought this hat seemed strange to the nurse. b.*The hat which that I brought seemed strange to the nurse was a fedora. The surface structure of'sentence (5.2.17) a is that shown in(5.2.18) (Ross 1967, p. 135). (5.2.18) 30 NP VP 3 /\ that NP VP A /\ I seemed I ,/’/’\\\ strange Y NP f) )3 brought to Afit I Art N ~ the nurse 146 In the derivation of sentence (5.2.17) b, the Sentential Subject Constraint given in (5.2.16) is violated, hence the sentence is ungrammatical. In contrast to this, in Marathi it is possible to relativize a noun phrase which is a constituent of the sentential subject, e.g. (5.2.19) a. tya manasa-ni tya 551-15 marala he the man-INSTR the woman-to beat this khara ahe true is 'That the man beat the woman is true.‘ b. tya mEnaSE-ni 51-15 marala he khara ahe the man-INSTR REL-to beat this true is ti bai ya gharat rahate that woman this house-in lives *‘The woman who that the man beat is true lives in this house.‘ The surface structure of sentence (5.2.19) a is that shown in (5.2.20). 14? (5.2.20) /So\ NP VP ha- k ara ahe true is he Si this tya manasa-ni the man-INSTR NP Y marala beat Art N I_ _I _ tya bai-la the woman-to Notice that sentence (5.2.19) b, which is perfectly gramma- tical, is derived by relativizing the NP tyE bai-la 'to the woman', a constituent of the sentential subject. The Coordinate Structure Constraint: ROSS (1967.pn 89) states this constraint as follows: (5.2.21) In a coordinate structure, no conjunct may be moved, nor may any element contained in a conjunct be moved, out of that conjunct. The following examples are from Ross (1967, pp. 88-89). (5.2.22) a. Henry plays the lute and sings madrigals. b.*The lute which Henry plays and sings madrigals 148 is warped. c.*The madrigals which Henry plays the lute and sings sound lousy. The surface structure of sentence (5.2.22) a is that shown in (5.2.23). (5.2.23) /S\ Nf’ VP Henry m VP and VP A A Y NP Y NF plays ’///\\\ sings madrigals Afit N the lute In the derivation of sentences (5.2.22) b and c, the Coordinate Structure Constraint is violated, hence these sentences are ungrammatical. More examples from English follow. (5.2.24) a. John and Mary went in. b.*Who did John and go in? (5.2.25) a. John bought the sofa and the chair. b.*What did John buy the sofa and? In contrast to this, in Marathi it is possible to either question or relativize an element contained in a conjunct, e.g. 149 (5.2.26) a. XSn-ni to SOpha' ani ti khurzi wikat anali John-INSTR the sofa and the chair bought 'John bought the sofa and the chair.‘ b. Kan-n1 to 80p 5 ani Xi khurCi wikat anali John-INSTR the sofa and REL chair bought ti khurei khup mahag ahe that chair very expensive is *‘The chair which John bought the sofa and is very expensive.‘ . h- - — . — c. Ken-n1 to 80p a ani kay w1kat anala? John-INSTR the sofa and what bought * 'What did John buy the sofa and?‘ As shown in sentences (5.2.26) b and c, the NP ti khurCi 'the chair', which is an element contained in the conjunct to sopha Eniti khurgi 'the sofa and the chair', may be either questioned or relativized, since neither of these processes involves the movement of that NP out of the conjunct. Notice that the English equivalents of these 2 sentences are ungrammatical. It may be concluded, therefore, that the Coordinate Structure Constraint given in (5.2.21) is not applicable in Marathi in the case of RELATIVE CLAUSE FORMATION and QUESTION FORMATION. The Pied Piping Convention: Ross (1967, p. 114) states this convention as follows: 150 (5.2.27) Any transformation which is stated in such a way as to effect the reordering of some specified node NP, where this node is preceded and followed by variables in the structural index of the rule, may apply to this NP or to any non-coordinate NP which dominates it, as long as there are no occurrences of any coordinate node, nor of the node S, on the branch connecting the higher node and the specified node. The following examples are from Ross (1967,p. 109). (5.2.28) a. The government prescribes the height of the lettering on the covers of the reports. b. Reports which the government prescribes the height of the lettering on the covers of are invariably boring. c. Reports the covers of which the government prescribes the height of the lettering on almost always put me to sleep. d. Reports the lettering on the covers of which the government prescribes the height of are a shock- ing waste of public funds. e. Reports the height of the lettering on the covers of which the government prescribes should be abolished. The surface structure of sentence (5.2.28) a is that shown 151 in (5.2.29) (Ross 1967, p. 109). (5.2.29) S NP VP the government V PrescribesP///’3\\\\P height P I of NP’///’/\\\\\‘NP the lettering the covers P ANP ' l of the reports The Pied Piping Convention given in (5.2.27), provides for the generation of all the relative clauses in sentences (5.2.28) b through e. Notice that RELATIVE CLAUSE FORMATION in Marathi does not require the Pied Piping Convention for the same reason as it does not require the Complex NP Constraint, the Sentential Subject Constraint and the Coordinate Structure Constraint: RELATIVE CLAUSE FORMATION is not a reordering transformation in Marathi.3 152 5.3 Focus and relative clause constructions in Marathi Schachter (1973) notes that in a variety of languages there are striking formal similarities between focus and restrictive relative clause constructions. The specific focus construction that Schachter refers to in his work is the cleft construction. According to Schachter ( p. 19) ”this suggests that there is some deep. non-language-specific relationship between the constructions [i.e. between focus and restrictive relative clause constructions] ." He gives the following examples from English to illustrate some of the morphological and syntactic similarities between the two constructions ( p.20): (5.3.1) a. It's Oedipus (who(m)/that) Jocasta loves. b. The man (who(m)/that) Jocasta loves. (5.3.2) a. It's Oedipus (which/that) Jocasta loves. b. The play (which/that) Jocasta loves. (5.3.3) a. It's Oedipus whose mother loves him. b. The man whose mother loves him. From the above examples it is clear that the relative pro- nouns that occur in both constructions involve the same human vs. non-human distinction and the same case marked forms. Also, the relative pronoun occurs in the same position in both types cf constructions and is deletable under the same conditions. The restrictive relative and the cleft constructions 153 in English are similar in another respect, namely. a noun phrase within a subordinate clause in both types of cone' structions can neither be relativized nor be made the focus of a cleft sentence, e.g. (Schachter 1973. p. 21): (5.3.h) a. This is the cat that chased the rat. b.*That's the rat that this is the cat that chased. I c.*It's the rat that this is the cat that chased. (5.3.5) a. It's the cat that chased the rat. b.*That's the rat that it's the cat that chased. c.*It's the rat that it's the cat that chased. Schachter (1973) claims that the above formal and syntactic similarities between the focus and relative clause constructions suggest that both of these constructions involve the process of promotion, that is. in both of them an element is promoted (i.e. moved) from an embedded clause into a main clause. in which it fills an originally empty slot. Further, Schachter prOposes that there is a semantic correlate for this syntactic prOperty of promotion which the two constructions have in common. This semantic correlate lies in the notion of foregrounding which he characterizes as follows (1973. p. 42). Fundamental to the focus-presupposition distinction is the division of the propositional meaning of a sentence into two parts. which are assigned different degrees of communicative importance. We may speak of this kind of division. in general terms, as the FOREGROUNDING of one part of the sentence at the expense of the other, and we may say that in focus constructions, the new information is 154 FOREGROUNDED, or IN THE FOREGROUND. while the presupposed information is BACKGROUNDED. or IN THE BACKGROUND. Schachter argues that as in focus constructions, in restrictive relative clause constructions there is also a ‘ foregrounding of one part at the eXpense of the other. That is, in restrictive relative clause constructions, the head NP, which determines the fUnction of that construction, is foregrounded at the expense of everything else in the relative clause. In contrast to English. in Marathi. relative clause and cleft constructions are not formally related. Before we compare these two types of constructions in Marathi, let us see how the cleft construction is formed in that language. Consider the following examples: (5.3.6) a. 35n-ni til-15' te pustak dila John-INSTR Bill-to the book gave 'John gave a book to Bill.’ b. bil-la te pustak kuni tar 35n-ni dila Bill-to the book who namely John-INSTR gave 'It was John who gave the book to Bill.‘ c. yen-n1 te pustak kunala tar bil-la dila John-INSTR the book who-to namely Bill-to gave 'It was Bill who John gave the book to.‘ d. Jon-n1 bil-la kay tar pustak dila John-INSTR Bill-to what namely book gave 'It was the book that John gave to Bill.‘ 155 Notice that the NPs 'John'. 'Bill' and 'book' are in focus in (5.3.6) a, b and c,respectively. As shown in these examples. the NP in focus is moved immediately in front of the verb, if it is not there already, and it is preceded by the following sequencexthe interrogative pronoun + £§£_ 'namely‘. thus separating that NP from any other NPs in the sentence which are not in focus. Note that in contrast to English, in Marathi the relative pronoun is formally distinct from the interrogative pronoun. For example, compare sentence (5.3.6) d with the following relative clause construction: (5.3.7) Yen-n1 bil-la' 3e dila te pustak khup John-INSTR Bill-to REL gave that book very mahag ahe expensive is 'The book which John gave to Bill is very expensive.‘ In sentence (5.3.6) d, which is a cleft construction, the interrogative pronoun kay 'what' is used. while in sentence (5.3.7). which is a relative clause construction, the rela- tive pronoun fig 'which' is used. Notice that in Marathi the relative pronoun makes the number and gender distinction but it does not make the human vs. non-human distinction. while the interrogative pronoun makes the distinction in the latter case but not in the former case, e.g. (5-3-8) JO REL 3e REL 31 REL fa? REL (5.3.9) a. 1.51 156 mulage.- / diwa- / ngh boy lamp tiger mulage / diwe /’we.1gh boys lamps tigers mulagi / wahi /’ sari girl notebookfem sarifem muli /'wahy§ / easy; girls notebooksfem sarisfem v -— kon tar Jan ala yesterday who namely John came 'It was John who came yesterday.‘ b. 1:31 kon tar meri ali yesterday who namely Mary came 'It was Mary who came yesterday.‘ v . - ~ .- hv- c. Jan-n1 kay tar to diwa p orala John-INSTR what namely the lamp broke 'It was the lamp that John broke.’ The relative clause and the focus constructions in Marathi differ from their English counterparts as follows: as mentioned earlier in this chapter, in English one can neither relativize a noun phrase from within a relative clause. nor can one get a focus construction with more than one noun phrase in focus. Sentences (5.3.h) and (5.3.5) above illustrate these two constraints in English. In contrast to this, in Marathi neither of these constraints is applicable. For, as pointed out earlier in this chapter, in Marathi it 157 is possible to relativize a noun phrase within the relative clause. Further. it is also possible to get a focus con- struction with more than one noun phrase in focus. e.g. (5.3.10) 35n-ni kuna-la tar bil-la kay tar John-INSTR whom namely Bill-to what namely pustak dila book gave 'John gave the book to Bill.‘ (the underlined NPs are in focus) (5.3.11) kuni tar gen-n1 kunala tar who namely John-INSTR whom namely bil—la kay tar pustak dila Bill-to what namely book gave 'John gave the book to Bill.‘ (the underlined NPs are in focus) Further, in Marathi it is possible to make a noun phrase within the relative clause the focus of the cleft con- struction, e.g. (5.3.12) 35n-ni He kunala tar bil-la dila John-INSTR REL whom namely Bill-to gave te pustak that book *‘The book which it was Bill that John gave to' In summary, the focus and relative clause constructions in Marathi are formally distinct. Further, these constructions in Marathi cannot be said to share the syntactic property of Promotion (Schachter, 1973). since the formation of a relative clause in Marathi does not involve the promotion of an NP 158 from an embedded clause into a main clause. Nevertheless, Schachter's notion of foregrounding neatly captures the semantic property shared by these two constructions. 5.4 Multiple headed relative clauses in Marathi In Marathi there are relative clauses with more than one head NP. It appears that Marathi has inherited these multiple headed relative clauses from Sanskrit, e.g. Andrews (1972, pp. 79-80): (5.4.1) a. yasyai yatj paitrkam riktham who-NOM what-NOM paternal-NOM inheritance-NOM sai tadj grahnita netarah he-NOM that-ACC should-get not another 'Of whom what is the paternal inheritance. he should get it and not somebody else.‘ b. yena yZth yatha 'dharma who-INSTR to-what-extent in-what-manner injustice dharma veha samihita sa eva justice or is done he exactly tatphalam bhfihkte tatha the-fruits-thereof will-enjoy in-that-way tavad amutra vai to-that-extent in-the-other-world indeed 'By whom good or evil is done to what extent in what way. he will enjoy the fruits of it in the other world to that extent in that way.‘ 159 Note that in the above Sanskrit sentences, which are instances of multiple headed relative clauses, we get a conditional paraphrase in the English translation. However, in Marathi we get multiple headed relative clauses with non-conditional interpretations as well. (5.4.2) jo manusi Ii jaminj kasato to REL man REL land tills that one i.e. he ticaj malak ahe that one-of i.e. its owner is 'Which man tills which land that man owns that land.‘ (5.4.3) 35' manasa-nii $5 bai-la marala REL man-INSTR REL woman-to beat-past to:L tica nawara. hot'a' that one i.e. he that one-of i.e. her husband was 'Which man beat which woman that man was that woman's husband.‘ Sentence (5.4.2) has a conditional interpretation. while sentence (5.4.3) has a non-conditional interpretation. Note that since English does not allow this type of relative clause. it is not possible to give an exact translation of the above Marathi sentences: the English equivalents are only approximations. Compare the following sentence with sentence (5.4.3): (5.4.4) 160 35' manasE—nii tya 631—155 marala toi REL man-INSTR the woman-to beat that one ticaj nawara hot; that one-of i.e. her husband was 'The man who beat the woman was her husband.’ Sentence (5.4.3) is a multiple headed (in this particular case, a double headed) relative clause, in that there is more than one head NP in the main clause of that sentence, whereas sentence (5.4.4) is a single headed relative clause. in that there is only one head NP in the main clause of that sentence. More examples of“multiple headed relative clauses in Marathi follow: (5.4.5) (5.4.6) 35' manasE-nii 35 bai-IEZ 3e pustakkdila hota REL man-INSTR REL woman-to REL book had given tinij tya-13.i te pustakk parat dila she-INSTR him that book returned 'Which man had given which book to which woman that woman returned that book to that man.' jii jacapeksa. unca hoti tya bai-nii REL REL-COMPARISON tall was that woman-INSTR tya manasE-sij lagna kela that man-with marriage did 'Which woman was taller than which man that woman married that man.' 161 (5.4.7) jahii 3185;15 lagna kela to minusi REL-INSTR REL-with marriage did that man butaka ahe Zni ti b'aij unca ahe short is and that woman tall is 'Which man married which woman that man is short and that woman is tall.‘ (5.4.8) 311 Xaéa-barobarj palun geli ti baii ani REL REL-with eloped that woman and to minusj ithe kém karit asat that man here work do used to 'Which woman eloped with which man that woman and that man used to work here.’ At this point the following question arises: Is there any semantic difference between single and multiple headed relative clauses in Marathi? There appears to exist some semantic difference between these two types of constructions. which may be explained as follows by using Schachter's (1973) notion of foregrounding:in single headed relative clauses only the head NP is foregrounded, while in double headed relative clauses both of the head NPs are foregrounded. Thus in sentence (5.4.4), which is a single headed relative clause, the head NP 'the man' is given prominence, while in sentence (5.4.3), which is a double headed relative clause. both of the head NPs (i.e. 'the man' and 'the woman') are given prominence. There seems to be some syntactic evidence to support this claim. Consider the following restrictive relative clauses in English. 162 (5.4.9) The man who beat the woman is the father of two children by her. (5.4.10) The woman whom the man beat is the mother of two children by him. Note that since in sentence (5.4.9) the NP 'the man' is fore- grounded, the NP 'the woman' cannot function as the subject in that sentence. Thus the following sentence is ungrammatical. (5.4.11) *The man who beat the woman is the mother of two children by him.‘ Likewise. since in sentence (5.4.10) the NP 'the woman' is foregrounded, the NP 'the man' cannot function as the subject in that sentence, e.g. (5.4.12) *The woman whom the man beat is the father of two children by her.‘ However. the proposal that in the double headed relative clauses in Marathi both of the head NPs are foregrounded predicts that in this type of clause in Marathi either of the two relativized NPs may function as the subject. In other words. this proposal Predicts that it is possible to get sentences in Marathi with the following readings. (5.4.13) a. 'Which man beat which woman he is the father of two children by her.‘ b. 163 'Which man beat which woman is the mother of two children by him.‘ And that is indeed the case. as the following double headed relative clause constructions in Marathi show: (5.4.14) (5.4.15) Furthermore, to the above (5.4.16) 55' manasa-ni 35 621-12 marala tyala REL man-INSTR REL woman-to beat to-him tizapasun don mula £311 her-by two children became 'Which man beat which woman he is the father of two children by her.‘ 35' manasZ-ni 35' 531-15 marala til; REL man—INSTR REL woman-to beat to-her tyacapasun don mula 231i him-by two children became 'Which man beat which woman she is the mother of two children by him.‘ certain cases of ambiguity also lend support claim. Consider the following sentence: to unca manusi tya} butakya manasa-jawalj that tall man that short man-near basal; hota 'ani tyaniij ty'alaij marala sitting was and he him beat 'The tall man was sitting near the short man and he beat him.‘ As shown in the above sentence, the pronominal forms tyani 161+ 'he' and ti§l§_'him' may refer either to unca manys 'the tall man' or butaka mégus 'the short man', hence the sentence is ambiguous. However. the relativization disambi- guates it, as shown in the following sentence: (5.4.17) jo unca manusi tya butakya manasa-jawa];j REL tall man the short man-near basal; hota tya'nii tyalaj marala sitting was he-INSTR him beat 'The tall man who was sitting near the short man beat him.‘ (5.4.18) 35' butakya' manasa'l-jawali to unca minusj REL short man-near the tall man basal; hota ty'a'nii tyalaj marala sitting was he-INSTR him beat 'The short man who the tall man was sitting near beat him.‘ Notice that in sentence (5.4.17) 'the tall man' is fore- grounded, while in sentence (5.4.18) 'the short man' is fore- grounded. However. if both of them are foregrounded (i.e. the double headed relative clause construction) the ambiguity creeps back in again, since in that case either of the NPs can function as the subject (i.e. the position of prominence): (5.4.19) jo unca minusi 35' butakya manasaLjawalj REL tall man REL short man-near basal; hota ty'a'niij tyalaij marala sitting was he-INSTR him ' . beat 'Which tall man was sitting near which short man he beat him.’ 165 (5.4.20) $5 butakya’ manasE-jawali jo unca minusj REL short man-near REL tall man basal; hota tyaniij tyalaij marala sitting was he-INSTR him beat 'Which tall man was sitting near which short man he beat him.‘ In each of the double headed relative clauses con- sidered thus far, one of the head NPs is the subject and the other is the object. In other words, in each of those cases the two head NPs do not have the same function. However, it is possible to get double headed relative clauses in Marathi, in which they have the same function. For example. consider the following sentence: (5.4.21) I saw the young man who was sitting near the young woman kiss her. In the above sentence the NP 'the young woman' is the object of the embedded clause, while the NP 'the young man', which is the subject of the embedded clause, is the object of the main clause. In other words. in sentence (5.4.21), both of the NPs. 'the young man' and 'the young woman' have the same function. It is possible in Marathi to get a double headed relative clause with both of the object NPs as the antecedents: 166 (5.4.22) mi joi ja- tarun b'éii-jawalj basal; hota tyE I REL REL young woman-near sitting was that tarun manasE-lai ticaj Bumban ghetana pahila young man-to her kiss taking saw 'Which young man was sitting which young woman I saw him kiss her.‘ To sum up, Schachter's (1973) notion of foregrounding neatly captures the semantic difference between single headed and multiple headed relative clauses in Marathi. As mentioned earlier in this study (i.e. section 4 of chapter 2), multiple headed relative clauses in Marathi seem to provide some internal evidence in support of the Conjunc- tion analysis for restrictive relative clauses. For consider the following sentence. (5.4.23) jewha jo mgnusi‘at'ala tewfia ji 6513 baher when REL man in came then REL woman out geli toi ani tij ekmekancfi dwes karatat went COREL and COREL each other'shate do 'When which man came in, which woman went out he and she (i.e. they) hate each other.’ The above sentence can be paraphrased in English only in the form of a conjunction. (This is true of the other double headed relative clause sentences given in this section.) Within the Conjunction analysis the underlying structure of sentence (5.4.23) may be represented as follows: 16? (5.4.24) (jewha minusi 3t 'ala' tewh; 5315 baher geli) when man in came then woman out went (manusi'ani 6313 ekmekanca dwes karatEt) man and woman each other's hate do In summary. section 1 describes the formation of questions in Marathi. In this section it is shown that in contrast to English. in Marathi it is possible to question NPs in relative clauses, with the exception of extraposed relative clauses. In section 2 it is shown that in contrast to English, in Marathi it is possible to relativize a noun phrase from within the embedded clause without violating such constraints by Ross (1967) as the Complex NP Constraint and the Sentential Subject Constraint. Further. in Marathi. but not in English, it is possible to either question or relativize a noun phrase which is contained in a conjunct, without violating Ross' (1967) Coordinate Structure Con- straint. Also. unlike English. Marathi does not require Ross' (1967) Pied Piping Convention. In other words. in the forma- tion of relative clauses and questions the Complex NP Constraint, the Coordinate Structure Constraint. the. Sentential Subject Constraint and the Pied Piping Convention are not applicable in Marathi. The reason for this is that in contrast to English, in Marathi RELATIVE CLAUSE FORMATION and QUESTION FORMATION are not reordering transformations. Furthermore. in section 3 it is shown that in contrast to English. in Marathi the relative clause construction is formally distinct from the cleft construction. Further. the 168 relative clause construction in Marathi does not lend support to the Promotion analysis (Schachter, 1973). since RELATIVE CLAUSE FORMATION is not a reordering transformation in this language. Next. in section 4 it is proposed that the semantic difference between single headed and multiple headed relative clauses in Marathi can be explained in terms of the notion of foregrounding (Schachter. 1973). Some syntactic evidence from Marathi is presented to support this claim. Lastly, it is suggested that multiple headed relative clauses in Marathi provide some internal evidence in support of the Conjunction analysis for restrictive relative clauses. CHAPTER 5 FOOTNOTES Kuno and Robinson (1972) give the following examples of echo and quiz questions in English: An echo question: Speaker A (almost inaudibly): "John expects that Mary will marry Bill." Speaker B:"What did you say? Who expects that Mary will marry who?" A guiz question : Quiz Show Moderator:"Now, for 500 dollars. who expects that Mary will marry who?” Dr. Yamuna Kachru (personal communication) finds sentence (5.2.26) b as ungrammatical. Sentence (5.2.26) c is acceptable to her only as an echo question. However, in my dialect (and I have verified it with other speakers of the dialect) sentence (5.2.26) b is perfectly gramma- tical. Further. sentence (5.2.26) c is acceptable as a genuine question, as the following discourse shows: 169 170 Speaker A:"kal amhi siyarsmadhe gelo hoto. sel yesterday we Sears-in went sale hot; tithe. kahi wastu khup swasta milalya. was there some things very cheap got hE sopha’ phakta tis dalaria miiala" this sofa only thirty dollars got 'Yesterday we went to Sears. There was a sale there. We got a few things very cheap (there). (For example,)we got this sofa for only $ 30. Speaker B:"kharac! sop ai'anakhi kay ghetala mag?" really sofa and what bought then *‘Really! what did you buy the sofa and?’ B's response is. in fact, a genuine question. and not an echo question. (There is no contrastive stress on fix 'what' in that question.) Like Marathi.in Hindi also it is possible to question an element contained in the conjunct (Keenan and Bimson. 1975). Japanese is similar to Marathi in this respect (KunO. 1973). Schachter's notion of foregrounding is similar in nature to that of theme/rheme, topic/comment and new/old information (Halliday (1967&1968), Kuno (1972), Novak and Sgall (1968). CHAPTER 6 CONCLUSION Listed below are the main conclusions of this study. (a) In this study the Conjunction analysis is assumed for restrictive and non-restrictive relative clauses in Marathi. Multiple-headed relative clauses in Marathi seem to provide some internal evidence from this language in support of the Conjunction analysis for restrictive relative clauses (see 5.4). (b) In Marathi. restrictive and non-restrictive relative clauses are not always formally distinct. Therefore. in the absence of any formal distinction between restrictive and non-restrictive relative clauses, their semantic inter- pretation is determined by the nature and extent of the hearer's knowledge about the referent of the head NP (see 3.4). (c) Some, if not all, non-restrictive relative clauses relate not only to the topic but also to the comment in the main clause. In such cases, the corresponding conjoin- ed sentences fail to express adverbial relations. such as 171 172 cause or concession (see sentences (3.4.9) b and (3.4.10) b). (d) In Marathi, relative clauses and relative participle phrases may be derived from the same underlying structure (see 4.1). Further, the formation of relative participle phrases in Marathi is subject to a constraint (i.e. (4.1.33)) which ensures that functional information is preserved in the surface structure of a sentence. In addition. it is shown that relative participle phrases in Marathi play a significant role in reducing the perceptual complexity of such structural configurations as self-embedding (see 4.2). (e) In the formation of relative clauses and questions. Ross' (1967) Complex NP Constraint, Coordinate Structure Constraint, Sentential Subject Constraint and Pied Piping Convention are not applicable in Marathi. The reason for this is that in contrast to English, in Marathi RELATIVE CLAUSE FORMATION and QUESTION FORMATION are not reordering transformations. (f) In contrast to English. in Marathi the focus and relative clause constructions are formally distinct (see 5.3). Nevertheless. they are similar to each other in one important respect: just as a focus construction can have more than one noun phrase in focus (e.g. (5.3.10) and (5.3.11)). so also a relative clause can have more than one relativized NP, i.e. multiple-headed relative clauses (e.g. (5.4.5) through (5.4.8)). This cannot be just an accidental 173 similarity: rather Schachter's (1973) notion of foreground- ing neatly captures the semantic property shared by these two constructions. (g) In Marathi a single formative. which begins with 3;. functions as a remote demonstrative, a third person pronoun and a correlative pronoun. If the remote demonstrative is taken as basic. it is possible to account for its use as a third person pronoun and correlative pronoun: it functions as a third person whenever the following noun has been deleted. Further, a correlative pronoun is of the same nature as the remote demonstrative in that it has deictic - and by extension anaphoric - properties: just as a remote demonstrative points to its referent, so the correlative pronoun points to the head NP in a relative clause. It is therefore proposed in this study that the presence of a correlative pronoun in a relative clause in Marathi serves as the surface structure clue that enables the hearer to relate the head NP to the preceding relative clause (see 3.2). Further. in contrast to English. in some cases Marathi makes a formal distinction between transparent and opaque contexts by means of the use of the correlative pronoun in the former and the adjective Egg 'such that: of this kind' in the latter (e.g. (3.2.11) a and b). (h) The following chart presents in summary form the transformational rules required for the derivation of restrictive and non-restrictive relative clauses as well as relative participle phrases: all three types of relative 174 constructions are derived.from the same underlying structure i.e. an unordered sequence of clauses. Restrictive Non- Relative relative restrictive participle clauses relative phrases clauses ANAPHORIC DEFINITIZATION + + + RCF + + + TRANSFER OF NOUN FEATURES + + + NP DELETION + + + RELATIVE PRONOUN DELETION + + LEFT DISLOCATION + + EXTRAPOSITION + + RIGHT DISLOCATION + + NP PREPOSING + ANAPHORIC DEFINITIZER DELETION + RELATIVE PARTICIPLE PHRASE FORMATION + 175 There are several important questions in regard to the main topic of this study which remain to be investigated. For example. the entire question of the quantified NPs containing relative clauses in Marathi remains unexplored. Also. there is no discussion in this study about such constructions in Marathi as infinitival and sentential relatives. BIBLIOGRAPHY Aissen. J. 1972. "Where do Relative Clauses Come from?" In J. Kimball (Ed.). Syntax and Semantics 1, pp. 187-198. New York and London: Seminar Press. Andrews, A. 1972. "A Typological Survey of Relative Clauses." MIT. MS Annear (Thompson). S. 1967. "Relative Clauses and Conjunctions." Working Papers in Lipguistics 1. Pp. 80-99. Ohio State University. . 1970a. "Relative Clause Structures and Constraints on Types of Complex Sentences." Workipg Papers in Lingpistics 6. pp. 20-40, Ohio State University. . 1970b. "The Deep Structure of Relative Clauses." Working Papers in Lingpistics 6, pp. 41-58, Ohio State University. Apte, M. 1962. A Sketch of Marathi Transformational Grammar. Unpublished Doctoral Dissertation, University of Wisconsin. Bach. E. 1968. "Nouns and Nounphrases." In E. Bach and R. T. Harms (Eds.). Universals in Linguistic~Theory. New York: 176 177 Holt, Rinehart and Winston. Baird, R. 1973. "Structural Characteristics of Clause-contain- ing Sentences and Imitation by Children and Adults." Journal of Psycholinguistic Research 12, pp. 115-127. Berman, A. 1974. "Infinitival Relative Constructions." Papers from the Tenth Regional Meeting of the Chicago Linguistic SOQie'LI: pp. 37-Ll'6o Bever, T. 1970. "The Cognitive Basis for Linguistic Structure." In J. R. Hays (Ed.), Cogpition and Language Learning, pp. 279-362. New York:Wiley. Bolinger, D. 1965. "Forms of English:Accent, Morphemep Order." Cambridge, Mass: Harvard University Press. Borkin, A. 1971. "Where do rules fail?" Indiana University Linguistics Club. Brame, M. 1968. "On the Nature of Relative Clauses." Unpublished paper. MIT. Carden, G. 1967. English Quantifiers. Unpublished M. A. thesis, Harvard University. Chafe, W. 1974. "Language and consciousness." Langpage 50, pp. 111-133. Chomsky, N. 1957. Syntactic Structures. The Hague: Mouton and Company. 178 . 1965. Agpects of the Theory_of Syntax." Cambridge, Mass: MIT Press. .1966. Cartesian Linguistics. New York: Harper and Row. Donaldson, S. 1971. "Movement in Relative Clauses in Hindi." In Y. Kachru (Ed.) Papers on Hindi Syntax, Studies in the Lingpistic Sciences (Working Papers) 1.2, pp. 1-74. University of Illinois. Drubig, B. 1968. "Some Remarks on Relative Clauses in English." Journal of English as a Second Languagg 3. pp. 23-40. Firbas, J. 1957. "Some Thoughts on the Function of Word Order in Old English and Modern English." Sbornik Praci V Filosofikefakulty Brneski UniyersityA5, pp. 72-94. Fodor, J. and M. Garrett. 1967. "Some syntactic determinants of Sentential complexity." Perception and Psychgphysics. 2, pp. 289-296. Fodor, J.. M. Garrett and T. Bever. 1968."Some syntactic determinants of Sentential complexity, II: verb strucutre." Perception and Psychophysics 3, pp. 453-461. Fodor, J.. T. Bever and M. Garrett. 1975. The Psychology of Langpage: an introduction to psycholingpistics and generative grammar. New York: McGraw Hill. Halliday.M.A.K. 1967-68. "Notes on transitivity and theme in English." Journal of;Linguistics 3 and 4. 179 Jackendoff, R. 1968. ”Speculations on Presentences and Determiners." Indiana University Linguistics Club. Jacobs, R. and P. Rosenbaum. 1968. English Transformational Grammar. Lexington, Mass: Xerox College Publishing. Junghare, I. 1973. "Restrictive Relative Clauses in Marathi." Indian Linguistics 34.4, pp. 251-262. Kachru. Y. 1973.""Some Aspects of Pronominalization and Rela- tive Clause Construction in Hindi-Urdu." In B. Kachru and Y. Kachru (Eds.), Papers pm South Asian Lingpistics, Studies_in the Lingpistic Sciences (Working Papers) 3.2, pp. 87-103. University of Illinois. Keenan, E. and B. Comrie. 1972. "Noun Phrase Accessibility and Universal Grammar.” Unpublished paper (presented at the Winter Meeting of the Linguistic Society of America). Keenan, E. and K. Bimson. 1975. "Perceptual Complexity and the Cross-Language Distribution of Relative Clause and NP-QUESTION Types." Papers from the Parasession on Functionalism, pp. 253-259. Chicago Linguistic Society. Kelkar, A. 1973. "Relative Clauses in Marathi: A Plea for a Less Constricted View." Indian Linguistics 34.4, pp. 27u-3000 Kimball, J.1973. "Seven principles of surface structure parsing in natural language." Cogpition 2, pp. 15-47. 180 Kuno, S. and J. Robinson. 1972. "Multiple Wh Questions." Lingpistic Inquiry 3.3, pp. 463-487. Kuno, S. 1972. "Functional Sentence Perspective: A Case study from Japanese and English." Lingpisticinquipy 3.2,: pp. 269-320. . 1973. The Structure oi_the Japanese Language. Cambridge, Mass: MIT Press. . 1975. "Three Perspective in the Functional Approach to Syntax." Papers from the Parasession on Functionalism, pp. 276-336, Chicago Linguistic Society. Kuroda, S. 1968. "English Relativization and Certain Related Problems." Langpage 44, pp. 244-266. Lakoff, G. 1968. ”Deep and Surface Grammar." Indiana Univer- sity Linguistic Club. Lambert, M. 1962. A Marathi Langpage Course. Bombay, India: Oxford University Press. Langendoen, D. 1969. The Study of Syntax. New York: Holt. Rinehart and Winston. Lees, R. 1963. The Grammar of English Nominalizations. Indiana University Research Center, Bloomington. Indiana. Loetscher, A. 1973. "On the Role of Nonrestrictive Relative Clauses in Discourse." Papers from the Ninth Regionai Meeting_of the Chicagnginguistic Socieiy, pp. 356-368. 181 Maling, J. 1972. "On 'Gapping and the Order of Constituents'." Linguistic Inquiry 3.1. pp. 101-108. Masica, C. 1972. "Relative Clauses in South Asia." The Chicago Which Hunt, Papers from the Relative Clause Festival. Chicago, Illinois: University of Chicago. McCawley, J. 1970. "Where Do Noun Phrases Come From?" In R. Jacobs and P. Rosenbaum (Eds.), Readings in English TransformationaiGrammar, pp. 166-188. Waltham, Mass: Ginn and Co. . 1974. "Pseudo-relative Clauses." Unpublished paper. Morgan, J. 1969. "On the Treatment of Presupposition in Transformational Grammar.“ Papers from the Fifth Regional Meeting_of the Chicago Lingaietic Society, pp. 167-177. . 1973. "How can you be in two places at once when you're not anywhere at all?" Papers from the Ninth Regional Meeting of the Chicago Linguistic Society, pp. 410-427. Chicago, Illinois: University of Chicago. . 1975. "Some remarks on the nature of sentences." Papers from the Parasession on Functionalism, pp. 433-449. Chicago Linguistic Society. Navalkar, G. 1925. The Student's Marathi Grammar. Poona, India: Scottish Mission Press. 182 Novak, P. and P. Sgall. 1968. "On the Prague Functional approach." Travaux Linguistique de Prague 3. Partee, ( Hall) B. 1972. "Opacity, Coreference, and Pronouns." In D. Davidson and G. Harman (Eds.), Semantics of Natural Languages. pp. 415-441. Holland: Reidel Dordrecht. Peranteau, P., J. Levi and G. Phares (Eds.), 1972. The Chicago Which Hunt, Papers from the Relative Clause Festival. Chicago, Illinois: University of Chicago. Postal, P. 1966. "On so-called 'pronouns' in English." In P. Francis and S. J. Dineen (Eds.), Report of the Seven- teenth Annual Round Tabie Meeting_on Lingeistics and Langgage Studies, pp. 177-206. Washington, D. C.: Georgetown University Press. . 1967. "Restrictive Relative Clauses and Other Matters." Unpublished paper. . 1971. Cross-over Phenomena. New York: Holt, Rinehart and Winston. Quine, W. 1960. Word and Object. Cambridge, Mass: MIT Press. Ross, J. 1967. Constraints_on Variables in Syntax. Unpublish- ed Doctoral Dissertation. MIT. . 1968. "Capping and the Order of Constituents." Indiana University Linguistics Club. 183 . 1970. "On Declarative Sentences." In R. Jacobs and P. Rosenbaum (Eds.), Readings in Engiish Transformational Grammar, pp. 222-272. Waltham, Mass: Ginn and Company. . 1973. "The Penthouse Principle and the order of Constituents." Papers from the Comparative Syntax Festival, pp. 397-422. Chicago, Illinois: University of Chicago. Ross, J. and D. Perlmutter. 1970. "Relative Clauses with Split Antecedents." Lingpistic Inquiry 1.3. p. 350. Scahchter, P. 1973. "Focus and Relativization." Langpage 49, pp. 19-46. Sheldon, A. 1974. "The Acquisition of Relative Clauses in English." Indiana University Linguistics Club. Slobin, D. and C. Welsh. 1973. "Elicited Imitation as a Research Tool in Developmental Psycholinguistics." In C. Fergusson and D. Slobin (Eds.), Studies in Child Language Develgpment, pp. 485-497. New York: Holt, Rinehart and Winston. Smith, C. 1964."Determiners and Relative Clauses." Langpage “'0’ pp‘ 37-52. Smith, M. 1974. "The Adequacy and Reality of Underlying Representations: Relative Clause Formation." Papers from the Tenth Regional Meeting of the Chicago Linguistic Society, pp. 643-656. Chicago, Illinois: University of Chicago. 184 Stockwell, R., P. Schachter and B. Partee (Hall). 1973. The Major Syntactic Structures of English. New York: Holt, Rinehart and Winston. Taglicht, J. 1972. "A New Look at English Relative Construc- tions." Lingua 29, pp. 1-22. Thorne, J. 1972. "On nonrestrictive relative clauses." Linguistic Inquipy 3.4, pp. 552-556- Ziv, Y. and P. Cole. 1974. "Relative Extraposition and the Scope of Definite Descriptions in Hebrew and English." Papers from the Tenth Regional Meeting of the Chicago Linguistic Society, pp. 772-785. Chicago, Illinois: University of Chicago. "'TITIl‘BIflITILflflQELEIfliifllflffli‘lfllifljflhflflfim'”