UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Word order variation in Japanese : characteristics of OSV word order Suzuki, Michiko 2000

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2000-0590.pdf [ 5.79MB ]
Metadata
JSON: 831-1.0099568.json
JSON-LD: 831-1.0099568-ld.json
RDF/XML (Pretty): 831-1.0099568-rdf.xml
RDF/JSON: 831-1.0099568-rdf.json
Turtle: 831-1.0099568-turtle.txt
N-Triples: 831-1.0099568-rdf-ntriples.txt
Original Record: 831-1.0099568-source.json
Full Text
831-1.0099568-fulltext.txt
Citation
831-1.0099568.ris

Full Text

WORD ORDER VARIATION IN JAPANESE: CHARACTERISTICS OF OSV WORD ORDER by MICHIKO SUZUKI B.A., Musashino Women's College, 1992 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS in THE FACULTY OF GRADUATE STUDIES (Department of Asian Studies) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA June 2000 © Michiko Suzuki, 2000 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of The University of British Columbia Vancouver, Canada DE-6 (2/88) Abstract This thesis attempts to explain the functions of the preposing of the direct object in Japanese. Japanese is generally a verb-final language and the basic word order is Subject (S) -> Object (O) -> Verb (V). However, it also has relatively free word order, and a direct object can be preposed to the sentence-initial position forming OSV word order. Although clauses expressed in SOV and OSV word orders are semantically identical, OSV is used much less frequently in comparison to SOV. This thesis examines the motivations behind the preposing of direct objects. As a first step toward understanding the possible functions of OSV word order, this thesis analyzes characteristics of preposed direct objects in its data collected from written materials. The results show that both structural characteristics and the information status of direct objects play roles in preposing. Structural characteristics found in preposed direct objects are that they tend to be lengthy, tend to contain demonstratives, or tend not be a direct object component of idiomatic expressions formed with a direct object and a verb. Preposed direct objects tend to convey information that is either linked to the preceding discourse, is linked to the hearer's knowledge, or is emphasized. The findings also show that the preposing of direct objects is motivated by various reasons depending on what is required to enhance communication in the context. Direct objects that have structurally different characteristics, namely those that contain demonstratives or long direct object phrases, are preposed for easier sentence processing or reference. In order to avoid ambiguity, preposing does not occur when the process would split an idiomatic expression comprised of a direct object and a verb. Preposing of direct objects conveying information linked to either the preceding discourse or the hearer's knowledge creates relevance between the preceding discourse and the present proposition. Preposing direct objects structurally indicates what information is emphasized. In other words, the preposing of direct objects facilitates effective communication. ii Table of Contents Abstract ii Table of Contents iii List of Tables v List of Figures vi List of Abbreviations vii List of Symbols viii Acknowledgements ix Chapter 1 Introduction 1 Chapter 2 Background of this Thesis 4 2.1 Syntactic Analyses of OSV Word Order 4 2.1.1 Mechanisms of Scrambling 6 2.1.2 Motivation of Scrambling 17 2.1.3 Summary 21 2.2 Pragmatic Analyses of Word Order Variations 22 2.2.1 Characteristics of Preposed Constituents 22 2.2.2 Information Status of Preposed Constituents 32 2.2.3 Summary 44 2.3 Hypotheses and Taxonomical Framework of this Thesis 44 2.3.1 Hypotheses of this Thesis 45 2.3.2 Terms and Taxonomy Employed in this Thesis 46 2.3.3 Summary 51 2.4 Summary 51 iii Chapter 3 Methodology 52 3.1 Data Sources 52 3.2 Data Collection 53 3.3 Data Analyses 60 Chapter 4 Analyses of Data 62 4.1 Distribution of SOV and OSV Clauses 62 4.1.1 SOV and OSV Word Orders in Three Types of Clauses 65 4.1.2 SOV and OSV Word Orders and Subjects Marked with Ga and Wa 68 4.2 Length of Direct Objects 71 4.3 Demonstratives in Direct Objects 78 4.4 Idiomatic Expressions 85 4.5 Emphasis 88 4.6 Information Status of Direct Objects 95 4.7 Summary 107 Chapter 5 Conclusion 109 5.1 Summary 109 5.2 Limitations 110 5.3 Suggestions for Further Study 111 References 112 Appendixes: Appendix A Data Sources 118 Appendix B List of Definitions 119 iv List of Tables 1. Summary of definitions of new and old information in previous studies 34 2. Information status of NPs (=Prince 1992: 309) 37 3. Definitions of information statuses 50 4. Distribution of SOV and OSV clauses 63 5. Distribution of SOV and OSV clauses in the seven source books 64 6. Distribution of main, subordinate, and noun-modifying clauses 67 7. Distribution of ga-marked and wa-marked subjects 70 8. Average number of bunsetsu units in S and O 71 9. Distribution of SOV and OSV clauses in relation to the number of bunsetsu units contained in S and O 72 10. Distribution of"+ Clause" and "- Clause" direct objects 77 11. Distribution of "+ Demonstrative" and "- Demonstrative" direct objects 79 12. Distribution of Type A and Type B "+ Demonstrative" direct objects 81 13. Distribution of the direct objects forming idiomatic expressions 86 14. Distribution of the emphasized direct objects in the OSV clauses 91 15. The information statuses of direct objects 100 v List of Figures 1. Assumed familiarity (= Prince 1981: 237) 35 2. Continuum of activation (= Dryer 1996: 482) 39 3. Idenfiability of information (= Lambrecht 1994: 109) 40 4. Information status of discourse entities: taxonomy of this thesis 49 vi List of Abbreviation ACC accusative case marker COMP complementizer CONT contrast marker CPL copula DAT dative marker EMPH emphasis GEN genitive marker LOC locative marker NC numeral classifier NMR nominalizer NOM nominative case marker NP noun phrase ONM onomatopoeia OSV Object (0) -> Subject (S) -> Verb (V) word order Q question marker QT quotative marker SFP sentence final particle SOV Subject (S) -> Object (0) -> Verb (V) word order TOP topic marker VP verb phrase vii List of Symbols Ungrammatical/unacceptable. Questionable. An increasing amount of question marks indicates increasing unacceptability. viii Acknowledgements I would like to express my appreciation to my thesis committee members, Dr. Yoko Collier-Sanuki, Dr. Henry Davis, Dr. Donald Baker, and Dr. Nam-Lin Hur, for their constructive comments and suggestions. Special gratitude goes to Dr. Yoko Collier-Sanuki, my graduate studies supervisor, for her constant encouragement and valuable feedback. I also would like to thank Dr. Henry Davis for his many insightful observations. I was fortunate to receive academic and moral support from many persons throughout my graduate studies. I am indebted to Allen Haaheim, Elizabeth Legge, and Eric Rosen for proofreading my thesis at various stages, and to my fellow graduate students in the Department of Asian Studies. The friends and families who showed me great kindness while living in Vancouver will always be warmly remembered. The office staff at Asian Studies was always ready to help me during my years at UBC and my appreciation for them is beyond words. Last but not least, I would like to thank my parents and sister in Japan for their unconditional support and patience. Though far away, knowing they were always there gave me constant motivation. Without their support I could not have completed this Master's of Arts degree. ix Chapter One Introduction Japanese is a verb-final language (Greenberg 1966) and the basic word order is Subject (S) -> Object (O) -> Verb (V). The order of sentence constituents other than verbs is relatively free; thus a simple transitive sentence 'John bought a book' can be expressed as follows: (1) a. Jon ga hon o katta.1 John NOM book ACC bought 'John bought a book.' b. Hon o jon ga katta. book ACC John NOM bought 'John bought a book.' The subject phrase Jon ga 'John NOM' and the direct object phrase hon o 'book A C C appear in the canonical order of SO in la, and in the reversed order of OS in lb. Although Japanese is a verb-final language, some constituents often appear following the verb in colloquial speech, and the same sentence 'John bought a book' can also be expressed as follows: (2) a. Hon o katta yo, jon ga.2 book ACC bought SFP John NOM (Lit.)'Bought a book, John.' 'John bought a book.' b. Jon ga katta yo, hon o. John NOM bought SFP book ACC (Lit.)'John bought, a book.' 'John bought a book.' The subject phrase jon ga 'John NOM' and the direct object phrase hon o 'book A C C appear following the verb and the sentence-final particle (SFP) katta yo in 2a and 2b, respectively. As examples 1 and 2 show, Japanese allows variation in the order of constituents. Because of such 'This thesis employs the Hepburn system of romanization. 2Sentence-final particles such as the soft confirmatory yo (Shibatani 1990) are often present in colloquial speech in Japanese. 1 word order variation, Japanese is considered to have relatively free word order. Many scholars have analyzed the word order variations shown in 2 and proposed that what appears post-verbally represents an "afterthought" (Martin, 1975; Shibatani, 1990), or that it functions to "repair" the preceding clause (Ono and Suzuki 1992). On the other hand, little research has been done with respect to the motivation behind the word order variation shown in 1. This thesis attempts to explain what motivates the direct object to precede the subject and appear in the sentence-initial position as in lb.3 Some previous studies propose that preposing in Japanese is due to some characteristics of certain sentence constituents, such as their length or their possession of demonstratives (Saeki 1960, 1975). It occurs when the information the constituents convey is to be emphasized (Kuno, 1973; Saeki, 1989), or is done in order to create a connection between the preceding discourse and the clause containing a preposed constituent (Masunaga 1983). However, these studies do not provide enough evidence from discourse to support their claims. The objective of this thesis is to verify or disprove their claims. Through its analysis of the characteristics of preposed direct objects, this thesis aims to understand the function of OSV word order in Japanese. There are several reasons to believe that OSV word order has certain functions distinct from that of SOV word order. First, the direct object is preposed in OSV clauses. Syntactic analyses of OSV clauses propose that an OSV clause is the result of movement of a direct object to a clause-initial position from its canonical position following the subject (cf. Saito 1985). Syntactic movement of constituents is generally taken to be motivated by syntactic, semantic, or pragmatic principle; for example, Heavy NP Shift in English, in which heavy or lengthy NPs are postposed, is motivated by focus considerations. Therefore, it is likely that the preposing of direct objects in Japanese is also motivated by linguistic principles. Second, OSV word order does not occur frequently compared to the basic SOV word order. According to Miyajima (1964), the ratio of 3In the Japanese language, both the direct object and the indirect object can precede the subject and create OSV word order. This thesis focuses on cases in which the direct object precedes the subject, and thus will use the abbreviation O in SOV and OSV to refer only to the direct object. 2 frequencies between SOV and OSV word orders is approximately 22.5 to l . 4 If a "speaker's choice of one syntactic form over another is not random" (Ward 1988), there must be a legitimate reason for a speaker to use OSV word order instead of SOV word order. The organization of this thesis is as follows: Chapter 2 first reviews previous studies of word order variations and preposing in Japanese, and then presents the hypotheses of this thesis. Chapter 3 describes the methodology used for the present research. Chapter 4 examines the hypotheses using the data collected for this study, following the methodology described in Chapter 3. Chapter 5 summarizes the findings of this study, and suggests issues for further study. 4Miyajima (1964) analyzes data collected from ninety magazines of various genres and found a total of 1365 clauses containing a subject, a direct object, and a verb. There are 1307 SOV clauses and 58 OSV clauses in his data. 3 Chapter Two Background of this Thesis This chapter reviews previous studies on Japanese word order variation in order to understand the nature of OSV word order. Section 2.1 reviews syntactic analyses that focus on the mechanisms of word order variation. Section 2.2 reviews pragmatic analyses that focus on the characteristics of preposed constituents and the function of preposing. Section 2.3 presents the hypotheses and framework of this thesis based on the review in Section 2.2. Section 2.4 summarizes Chapter 2. 2.1 Syntactic Analyses of OSV Word Order The phenomenon of optional word order variation is referred to as "scrambling" in syntax, and sentences in which some constituents are not in canonical order, such as lb below, are called "scrambled" sentences. (1) a. Canonical word order Jon ga hon o katta. John NOM book ACC bought 'John bought a book.' b. "Scrambled" sentence Hon o jon ga katta. book ACC John NOM bought 'John bought a book.' There are three types of scrambling: a) clause-internal scrambling, in which a constituent other than a subject appears in the clause-initial position; b) verb phrase (VP)-internal scrambling, in which the direct object appears to the left of the indirect object within the VP; and c) long-distance scrambling, in which a constituent in an embedded clause appears in the sentence-initial position.1 'The following examples (i) and (ii) demonstrate VP-internal scrambling and long-distance scrambling respectively. The underlined direct object phrases appear in canonical order in the "a" sentences and in scrambled positions in the "b" sentences. 4 The OSV sentence in lb, in which the direct object phrase hon o 'book A C C appears in the sentence-initial position, is an instance of clause-internal scrambling.2 In the following subsections, I will review analyses of clause-internal scrambling. Section 2.1.1 reviews analyses of how scrambled constituents appear in the clause-initial position. Section 2.1.2 reviews analyses of motivations for scrambling. Section 2.1.3 summarizes the syntactic analyses reviewed in Section 2.1. (i) VP-internal scrambling a. Jon ga marii ni hon o ageta. John N O M Mary D A T book A C C gave 'John gave Mary a book.' b. Jon ga hon o marii ni ageta. John N O M book A C C Mary D A T gave 'John gave Mary a book.' (ii) Long-distance scrambling a. Jon ga marii ga hon o katta to itta. John N O M Mary N O M book A C C bought QT said 'John said that Mary bought a book.' b. Hon o jon ga marii ga katta to itta. book A C C John N O M Mary N O M bought QT said 'John said that Mary bought a book.' Long-distance scrambling shares a characteristic with clause-internal scrambling in that the constituents are moved to A'-position (Section 2.1.1 reviews analyses of the positions constituents appear in clause-internal scrambling). However, this thesis will not discuss the long-distance scrambling of direct objects as shown in (iib) because long-distance scrambling does not occur frequently in speech or writing. For further analyses of VP-internal scrambling and long-distance scrambling in the Japanese language, see Saito (1985, 1992) and Miyagawa (1997). 2The scrambling of constituents other than direct objects, such as indirect objects, adverbs, etc., can also occur in clause-internal scrambling. Also, clause-internal scrambling of more than one constituent (multiple scrambling) is possible. Both the indirect object and the direct object are scrambled in the example below: Marii ni hon o jon ga ageta. Mary D A T book A C C John N O M gave 'John gave Mary a book.' Since this study focuses on the clause-internal scrambling of direct objects, I will not provide or discuss examples of the scrambling of constituents other than direct objects and multiple scrambling. 5 2.1.1 Mechanisms of scrambling There are two proposals on how a scrambled sentence is created: a) the direct object in OSV sentences originates (i.e. is base-generated) in the sentence-initial position without movement (cf. Hale, 1980); and b) it is moved to the sentence-initial position from its canonical position (cf. Harada, 1977; Saito, 1985). This section will review these two proposals. Hale (1980) proposes that arguments of a verb, such as the subject and direct object of a transitive verb, can appear in any order with reference to each other. In other words, the direct object in OSV sentences is base-generated in the sentence-initial position. His proposal is based on the following syntactic structure: (2) VP NP NP NP V The above structure shows VP with a single level of structure, with the V, which is the head of VP, appearing in the final position. The NPs have the same hierarchical status within the VP (=nonconfigurational structure), and the order in which they appear in the VP is free. The SOV sentence la would be described as in 3 following Hale's proposal: (3) VP NP NP V Jon ga hon o katta John NOM book ACC bought 'John bought a book.' The verb katta 'bought' in 3, which is the head of the VP, must be sentence-final. On the other hand, the NPs Jon ga 'John NOM' and hon o 'book A C C can appear in any position as long as they precede the verb katta 'bought.' Since the order of the NPs within the VP is free in Japanese, the sentence 'John bought a book' can be Jon ga hon o katta, as in la, or Hon o jon ga katta, as in lb. In other words, Hale proposes that lb is not derived from la by preposing the direct object phrase hon o 'book A C C to the sentence-initial position. 6 Saito (1985) argues against Hale's nonconfigurational analysis of Japanese and proposes that scrambling in Japanese is a movement operation. He proposes that Japanese is not a nonconfigurational language as in 3 but a configurational language like English, in which the subject and object have different hierarchical statuses. The structure below represents his proposal: (4). S NP VP (Subject) NP V (Object) In the above representation the first branching node dominating the subject NP also dominates the object NP, but the subject NP does not dominate the object NP. This relationship between the two constituents is called "c-command," and is defined by Reinhart (1979) as follows: (5) c-command Node A c-commands node B if and only if (i) A does not dominate B and B does not dominate A; and (ii) the first branching node dominating A also dominates B. (Reinhart 1979: 114) Thus, the subject c-commands the object but the object does not c-command the subject. In other words, the hierarchical status of the subject and the object is asymmetrical. The structure in 6 below describes the SOV sentence la, Jon ga hon o katta 'John bought a book,' based on Saito's proposition that Japanese is a configurational language. (6) S NP VP Jon ga John NOM hon o katta book ACC bought 'John bought a book.' The subject NP and the direct object NP have different hierarchical statuses in the structure in 6; 7 they are not dominated by all the same nodes and the subject NP c-commands the direct object NP. Since the direct object NP is lower in the tree structure than the subject, the direct object has to actually move to a sentence-initial position to precede the subject and create an OSV sentence. Saito provides three pieces of evidence for his argument against Hale's analysis. The first is based on pronominal coreference. A pronoun and its antecedent generally hold the relationship described in 7.3 (7) A pronoun cannot c-command its antecedent. Saito presents the following English paradigm of sentences containing pronouns and their antecedents.4 . (8) a. [s John. [ y p loves [Np his. mother ]]5 b. *L He. [.„ loves L,n John's mother 111 LS i L VP LNP i J J J c [c LTD John.'s 'mother [,,„ loves him. ]] LS LNP l LVP l J J d. [s [ N p H i S j mother] [ loves John. ]] (Saito 1985: 36) Sentence 8b is ungrammatical because the pronoun 'he,' which is the subject, c-commands its antecedent 'John' in the direct object NP, while the pronouns in 8a, 8c, and 8d do not. If English is a nonconfigurational language with the structure represented in 5 and if it lacks a VP, as Hale proposes for Japanese, the pronoun in 8c c-commands its antecedent and the sentence should be classified as ungrammatical. In contrast, 8c is grammatical. Saito uses paradigm 8 to demonstrate that the object does not c-command the subject and that there is a hierarchical relationship between 3Saito (1985) states that condition 7 can be considered as part of the following more general principle suggested in Higginbotham (1983): If X c-commands Y, then Y is not an antecedent of X. (Higginbotham 1983: 402) 4Coindexation indicates that a pronoun and its antecedent in a sentence refer to the same individual, and an asterisk indicates that the sentence is ungrammatical. 5From example 8 on, I will use square brackets to represent syntactic structure instead of tree structure. Each square bracket indicates phrases. 8 the subject and the object. Saito refers to Whitman's (1982) paradigm in Japanese which shows the same result as 8 with respect to the pronominal coreference as shown below:6 (9) a. Jori[ ga [NP marii ga kare\ ni okutta tegami ] o John NOM Mary NOM he DAT sent letter ACC mada yondeinai (koto) yet have-not-read fact 'John, has not read the letter Mary sent to him..' b. *Kare\ ga [NP marii ga jon\ ni okutta tegami ] o he NOM Mary NOM John DAT sent letter ACC mada yondeinai (koto) yet. have-not-read fact * 'He. has not read the letter Mary sent to John..' c. [NP Jon\ kara okane o moratta hito ] ga karex o John from money ACC received person NOM he ACC suisenshita (koto) recommended fact 'The person who received money from John, recommended hirn.' d. [NP Karex kara okane o moratta hito ] ga jonx o he from money ACC received person NOM John ACC suisenshita (koto) recommended fact 'The person who received money from himj recommended John;.' (Whitman 1982) (cited in Saito 1985: 37-38) If Japanese lacked a VP, 9c would be ungrammatical—unlike its English counterpart 8c. The pronoun kare 'he' in the direct object NP would c-command its antecedent jon 'John' in the subject NP and violate condition 7. Paradigm 9 provides some evidence that Japanese has a configurational structure and that its pronouns are also subject to condition 7. 6Saito (1985) adds koto 'the fact that' at the end of some examples to avoid the unnaturalness that results from the lack of a topic in a main clause. However, he does not translate koto. 9 Saito employs pronominal coreference to further demonstrate that the direct object does not c-command the subject, even when it precedes it. This is shown in 10: (10) a. *Kare\ ga [ N P marii ga jonx ni okutta tegami ] o he NOM Mary NOM John DAT sent letter ACC mada yondeinai (koto) yet have-not-read fact * 'He; has not read the letter Mary sent to John..' b. [NP Marii ga jon\ ni okutta tegami ] o kare\ ga Mary NOM John DAT sent letter ACC he NOM mada yondeinai (koto) yet have-not-read fact 'The letter that Mary sent to John., he; has not read.' (Saito 1985: 39-40) The pronoun kare 'he' precedes its antecedent jon 'John' in the direct object NP; thus sentence 10a violates condition 7. On the other hand, in 10b, the antecedent jon 'John' in the scrambled direct object NP precedes the pronoun kare 'he'; nevertheless, this sentence is grammatical. Saito argues that if we assume that Japanese is a nonconfigurational language as in 3, where a direct object NP can appear in any position as Hale assumes, the pronoun kare 'he' would c-command its antecedent jon 'John' and the sentence would be ungrammatical. The grammaticality of 10b supports Saito's claim that the scrambled direct object NP in 10b has been moved to a position in which the pronoun in the subject position does not c-command it. Second, Saito demonstrates that scrambled constituents are moved, not base-generated,- on the basis of the process known as "quantifier stranding." A quantifier normally appears preceding its modifying NP; however, it is "stranded" when the modifying NP appears preceding the quantifier. Examples 11 and 12 from Saito (1985) show quantifier stranding from a subject and from a direct object, respectively:7 7 A quantifier can be stranded only from a subject NP or a direct object NP (Shibatani 1990). The following examples show that stranding a quantifier from an indirect object NP or an oblique NP is impossible. (i) A quantifier stranding from an indirect object NP: 10 (11) a. San-nin no gakusee ga sake o nondeiru. three-NC GEN student NOM sake ACC drinking 'Three students are drinking sake.' b. Gakusee ga san-nin sake o nondeiru. student NOM three-NC sake ACC drinking 'Three students are drinking sake.' (Saito 1985: 51) (12) a. Jon ga san-bon no sake o mottekita. John NOM three-NC GEN sake ACC came-with 'John came with three bottles of sake.' b. Jon ga sake o san-bon mottekita. John NOM sake ACC three-NC came-with 'John came with three bottles of sake.' (Saito 1985: 51) Quantifiers modifying NPs are marked by the genitive case marker no as in 11a and 12a. When the quantifiers are stranded, they are not marked by no and follow the NP they modify as in 1 lb and 12b. Scrambling the direct object NP in sentence 12b, and leaving the quantifier stranded, is possible as shown in 13. a. Boku wa san-nin no kodomotachi ni hon o yatta. I TOP three-NC G E N children D A T book A C C gave 'I gave books to three children.' b. *Boku wa kodomotachi ni san-nin hon o yatta. I TOP children D A T three-NC book A C C gave T gave books to three children.' (Shibatani 1990: 286) (ii) A quantifier stranding from an oblique NP: a. Boku wa san-nin no kodomotachi kara hon o moratta. I TOP three-NC G E N children from book A C C got 'I got books from three children.' b. *Boku wa kodomotachi kara san-nin hon o moratta. I . TOP children from three-NC book A C C got T got books from three children.' (Shibatani 1990: 286) 11 (13) Sake o jon ga san-bon mottekita. sake ACC John NOM three-NC came-with 'John came with three bottles of sake.' (Saito 1985: 52) Although the subject NP jon ga 'John NOM' intervenes between the scrambled direct object and its stranded quantifier, 13 is grammatical and the stranded quantifier is interpreted as modifying the direct object NP. On the other hand, a quantifier cannot be interpreted as modifying the subject when the direct object intervenes between them, as shown in the following example from Saito: (14) *Gakusee ga sake o san-nin nondeiru. student NOM sake ACC three-NC drinking 'Three students are drinking sake.' (Saito 1985: 52) Using the contrast between 13 and 14, Saito argues that SOV in 12b is the basic word order and OSV in 13 is derived from 12b via a movement of the direct object. He proposes that 13 has the following syntactic representation: (15) [s Sake o\ [s jon ga t\ san-bon mottekita ] ] sake ACC John NOM three-NC came-with 'John came with three bottles of sake.' (Saito 1985: 53) The preposed direct object NP Sake o 'sake A C C leaves a trace (t) in its original position, and the stranded quantifier can be associated with the direct object indirectly via this trace. Third, Saito claims that scrambling is sensitive to the crossover effect. Syntactic analyses distinguish two types of positions: A-positions and A'-positions. A-positions are positions to which a theta role such as an agent, patient, theme, etc. can be assigned. A'-positions are positions which are not A-positions. Crossover effect is observed when a quantifier phrase or wh-phrase is moved to an A'-position. However, the examples Saito provides to illustrate crossover effect involve neither quantifier phrases nor wh-phrases, as shown below: (16) a. [ N P / C W J no hahaoya ] ga karex o aishiteiru (koto) John GEN mother NOM he ACC love fact 'John.'s mother loves him..' i i 8See Lasnik and Stowell (1991) for further discussion of crossover effect. 12 (16) b. ? *[NP^0ftj no hahaoya ] o kare\ ga aishiteiru (koto)9 John GEN mother ACC he NOM love fact ??/?*'John.'s mother, he. loves.' (Saito 1985: 48) Saito argues as follows: A subject pronoun can have its antecedent within a scrambled direct object that precedes it, as in example 10b. On the other hand, example 16b above shows the opposite result. The direct object NP is moved to the sentence-initial position in 16b; thus, the pronoun kare 'he' in the subject position does not c-command its antecedent Jon 'John' as in 10b. However, 16b is ungrammatical or questionable while 10b is grammatical. Saito claims that 16b is an instance of crossover and provides the generalization in 17. (17) When the object precedes the subject, a pronoun in the subject position can take a name contained in the object as its antecedent only if the name is embedded "deeply enough" within the object. (Saito 1985: 48) The above generalization explains the gramrnaticality and ungrammaticality of the scrambled sentences 10b and 16b respectively. Saito argues that 10b is grammatical because the antecedent jon 'John' is embedded "deeply enough" within the preposed object NP while the antecedent in 16b is not. He claims that the crossover effect observed in 16b is found only when an element containing an antecedent is moved to an A'-position across a coindexed pronoun, and suggests that the direct object in 16b has been moved to an A'-position based on the analysis of 16b. He then argues that Hale's (1980) nonconfigurational analysis does not provide an explanation for the difference in grammaticality between 10b and 16b because, under his analysis, 10b and 16b have exactly the same structural representations, and the pronoun kare 'he' c-commands its antecedent jon 'John' in both structures. The scrambling of a direct object can be either to an A'-position or to an A-position (Saito and Hoji, 1983; Saito, 1985, 1992). Saito (1985, 1992) demonstrates an instance of scrambling to 9 A question mark indicates that the sentence is questionable. An increasing amount of question marks indicates increasing unacceptability. 13 an A'-position, shown in the example below, employing anaphor binding. (18) a. [s Hanako\ ga jibunjishin^ o hihanshita] (kotp)^ Hanako NOM self ACC criticized fact 'Hanako criticized herself.' b. [s Jibunjishin^ o [s hanakox ga [yp tj hihanshita ]]] (koto) self ACC Hanako NOM criticized fact 'Herself;, H a n a k O j criticized.' (Saito 1992: 76) Jibunjishin 'self is an anaphor and it must be c-commanded by a coindexed NP in the A-position for the sentence to be grammatical. This relationship between an anaphor and a coindexed NP is described in condition 19 below: (19) An anaphor must be A-bound by its antecedent. The anaphor jibunjishin 'self is A-bound by its antecedent Hanako 'Hanako' in 18a so the sentence satisfies condition 19. On the other hand, the antecedent hanako 'Hanako' is a referential expression and cannot be c-commanded by a coindexed NP. This is described in condition 20 below: (20) Referential expressions must be A-free. Example 18a satisfies both condition 19 and 20. In example 18b, the direct object NP containing the anaphor jibunjishin 'self is moved to the sentence-initial position and c-commands its antecedent hanako 'Hanako.' Saito argues that the anaphor jibunjishin must be in the A'-position in 18b because the sentence would violate condition 20, and thus be ungrammatical if the anaphor were in the A-position. Saito and Hoji (1983) and Saito (1985) demonstrate that scrambling can be to an A'-position by employing the diagnostic test of weak crossover. Weak crossover is observed when a 1 0The following is the definition of binding: A binds B if and only if (i) A c-commands B; and (ii) A and B are coindexed. 1 'Example 18a is my addition. The anaphor jibunjishin and jibun 'self are usually spelled zibunzishin and zibun respectively in syntactic analyses. 14 quantifier phrase or wh-phrase is moved to an A'-position across a pronoun with the same reference.12 However, the examples Saito and Hoji (1983) and Saito (1985) provide to demonstrate the environment of "weak crossover" differ from the environment in which weak crossover effect is generally observed in that their examples involve movement of neither quantifier phrases nor wh-phrases. The second type of scrambling, that of constituents to the A-position, is shown in the example below from Saito (1992): (21) a. Masao ga [VP [NP otagai\ no sensee] ni karera\ o .Masao NOM each other GEN teacher to they . ACC shookaishita ]] (koto) introduced fact ? * 'Masao introduced them to each other's teachers.' (Saito 1992: 74) b. [s Karera\ o [s masao ga [vp [NP otagai\ no sensee ] ni t\ they ACC Masao NOM each other GEN teacher to syookaishita ]]] (koto) introduced fact 'Masao introduced them to each other's teachers.' (Saito 1992: 75) The anaphor otagai 'each other' is not A-bound by the coindexed antecedent karera 'they' in 21a, so this sentence is ruled out by condition 19. However, it is considered grammatical when the object NP containing the antecedent karera 'they' is moved to a sentence-initial position as in 21b. Saito argues that 21b is grammatical because the object NP is moved to an A-position and the 1 ^ antecedent karera 'they' A-binds the anaphor otagai 'each other,' thus satisfying condition 19. 1 2See Lasnik and Stowell (1991) for further discussion of weak crossover effect. l3Long-distance scrambling is also A'-movement as shown below: *[ Karera\ o [$ otagai\ no sensee ga [ [s hanako ga tj hihanshita ] to ] they A C C each other G E N teacher N O M Hanako N O M criticized QT itta ]] {koto) said fact 'Theirij, each other'Sj teachers said that Hanako criticized t\ (Saito 1992: 70) Saito (1992) further explains a crucial difference between scrambling to an A-position and an A'-position. A-scrambled constituents are interpreted in their moved position and A'-scrambled constituents in their trace position at a level called Logical Form (LF). 1 4 Therefore, the A-scrambled pronoun karera 'they' in example 21b remains in the sentence-initial position at LF, and is thus interpreted as an A-binder of the anaphor otagai 'each other.' On the other hand, A'-scrambling of the anaphor jibunjishin 'self in 18b is undone at LF, and jibunjishin 'self is interpreted as having hanako 'Hanako' as its antecedent.15 Hale's (1980) analysis, which proposes that scrambling does not involve movement, is not able explain how the A'-scrambled anaphor jibunjishin 'self can be given this interpretation. This section has reviewed two types of analyses of scrambling; base-generation and movement. Through the review of Saito's (1985) analyses, we have seen that Japanese is a configurational language, and OSV sentences are created by moving the direct object to the sentence-initial position. We have also seen that scrambling can be to two types of positions; A-position and A'-position (Saito and Hoji, 1983; Saito, 1985, 1992). Based on the review in this section, this thesis follows Saito's analyses and assumes that the direct object in OSV clauses has been moved to the clause-initial position, or "preposed." Although we now understand how a scrambled direct object appears in the sentence-initial position, what motivates scrambling is not yet clear.16 Movement of constituents generally occurs either in order to receive a certain kind of feature, or in order not to receive a certain kind of feature. For example, wh-phrases in English move to get their wh-feature checked. A direct object in Dutch moves away from its original position, which is a focus position, in order to be l 4 L F is a syntactic level of representation which encodes logico-semantic properties of a sentence. 1 5See Saito (1989, 1992) for more detailed analyses of L F status of scrambling. l 6Recent syntactic analyses explain scrambling in light of Chomsky's (1992) Minimalist Program. However, they too do not explain the motivation of scrambling. For further analyses of scrambling under the Minimalist Program, see Saito and Fukui (1998) and Boskovic and Takahashi (1998). 16 defocused (Reinhart, 1995). The following section will review a syntactic analysis which attempts to explain why some constituents are scrambled in Japanese. 2.1.2 Motivation of Scrambling Miyagawa (1997) proposes that there are two motivations for scrambling constituents, and as a result, we have scrambling to A- and A'-positions. First, he proposes that scrambling to an A-position is motivated by "some sort of agreement feature" (1997: 3). He suggests that the functional head "I," where nominative and accusative cases are checked, must be unpacked into a more articulated set of functional heads, namely subject agreement (Agrs) for nominative case checking, and object agreement (Agro) for accusative case checking. He claims that in Japanese Agro is inherently weak but that Agrs is strong. Since Agro is weak, it is not until LF that the accusative case of the object is checked in the specifier position of the Agro to the right of subject. 11 He demonstrates this in the following example: (22) [ip Subject NOM Object A C C Agro Agrs ] t T- check A C C case 1 check NOM case (Miyagawa 1997: 15) Miyagawa further claims that Agro can be incorporated into Agrs. When Agro fuses with Agrs, the fused head takes on the strong feature of Agrs, which is the head of the newly created category. In this instance, the accusative case of the object can be checked in the IP-adjoined position at the overt syntax as shown in 23: (23) [IP Object A C C [IP Subject NOM Agro-Agrs ]] ^ 1- check A C C case 1 '— check NOM case ' (Miyagawa 1997: 15) He proposes that 23 represents an environment in the which scrambling of a direct object to sentence-initial position is possible. In other words, Agrs must be present for an A-scrambling. 1 7 In earlier syntactic theories, clauses were marked by S, but in later theories, a sentence is considered to be a projection of a verbal inflection head labelled INFL. Since a sentence is a phrase whose head is INFL, IP is used to mark clauses instead of S. Miyagawa employs this IP instead of S to mark clauses. 17 He provides example 24 to further explain the difference created by the presence or absence of Agrs. We have seen in the previous section that an anaphor must be A-bound by its antecedent for a sentence to be grammatical. (24) a. [NP [IP Karera\ o [\p otagaix no sensee ga tj kiratteiru ]] riyuu ] they ACC each other GEN teacher NOM hate reason 'the reason that them;, each other's teachers hate' b. ? ? ?/*[NP [IP Karera\ o [IP otagai\ no sensee no tj they ACC each other GEN teacher GEN kiratteiru ]] riyuu ] hate reason 'the reason that them., each other's teachers hate' (Miyagawa 1997: 18) The anaphor otagai 'each other' is A-bound by its antecedent karera 'they,' which is in the scrambled direct object NP; thus 24a is grammatical. The anaphor otagai 'each other' is also A-bound by its antecedent in 24b; however, it is ungrammatical. Miyagawa argues that the difference in grammaticality or acceptability between 24a and 24b is due to the presence or absence of Agrs. The nominative case marker ga in 24a indicates the presence of Agrs; thus, the scrambling of the object NP karera o 'they A C C to an A-position is possible. On the other hand, Agrs is not present in 24b and the subject sensee 'teacher' is marked by the genitive case marker no. Therefore, scrambling the direct object NP to an A-position in 24b results in an ungrammatical or unacceptable sentence. Second, Miyagawa proposes that scrambling to an A'-position is motivated by "focus" (1997: 3). He explains that A'-scrambling is possible where A-movement is not required for the purpose of providing an A-binder, and that the construction sounds best if the A'-moved constituent receives a heavy focus accent. He provides examples which contain direct object NPs marked by the topic/contrast marker wa as shown in 25.18 Words in capital letters in his 1 8Miyagawa (1997) also provides examples in which A'-moved direct objects marked by o to support his proposal that A'-scrambling is motivated by focus. The nominative case marker ga may be replaced by the genitive case marker no in noun modification as shown in (i): 18 translation indicate that they convey focus. (25) a. ^Jon ga isoide hon wa katta. John NOM quickly book CONT bought 'John quickly bought A BOOK.' b. Hon wa\ jon ga isoide tj katta. book CONT John NOM quickly bought 'John quickly bought A BOOK.' (Miyagawa 1997:22) He suggests that 25b sounds better when the constituent marked by the contrastive marker wa is moved than when it remains in its original position as in 25a. Therefore, he proposes that scrambling to an A'-position is movement to a focus position. I found that there are several problems with Miyagawa's analysis. He argues that A-scrambling is possible only when Agrs is present, but that A'-scrambling is possible when Agrs is either present or absent. He demonstrates A'-scrambling in the absence of Agrs in the following example. The use of bold face indicates the focus accent in his example. (i) [NP tip Tanaka no piza o tabeta ] ] mise ] Tanaka G E N pizza A C C ate restaurant 'the restaurant where Tanaka had pizza' Based on his analysis, Agrs is not present in the above example; thus, the direct object can only undergo A ' -movement. He argues that the sentence sounds best when the A'-moved direct object receives heavy focus as in (iii) but that it is less acceptable without a heavy focus accent as in (ii). Bold face indicates focus accent in his examples. (ii) ??1NP [IP piza 0[ [\y tanaka no tj tabeta ] ] mise ] pizza A C C Tanaka G E N ate restaurant 'the restaurant where Tanaka had pizza' (iii) [NP [IP piza °\ [NP tanaka no tj tabeta ] ] mise ] pizza A C C Tanaka G E N ate restaurant 'the restaurant where Tanaka had pizza' (Miyagawa 1997: 22) (Miyagawa 1997: 22) 1 9 The capitalization is from Miyagawa's original. 19 (26) a. [NP [IP Tanaka no piza o tabeta ] ] mise ]20 Tanaka GEN pizza ACC ate restaurant 'the restaurant where Tanaka had pizza' b. [NP [IP piza ox [NP tanaka no tx tabeta ] ] mise ] pizza ACC Tanaka GEN ate restaurant 'the restaurant where Tanaka had pizza' (Miyagawa 1997: 22) The direct object NP piza o 'pizza A C C is moved to an A'-position in 26b. Although Miyagawa claims that 26b is acceptable, it is a very unnatural sentence even with the A'-moved direct object receiving a heavy focus accent. We have seen in the previous section that an anaphor jibunjishin 'self is moved to an A'-position in a grammatical scrambled sentence. An example of A'-scrambled jibunjishin 'self from Saito (1992) is repeated in 27: (27) [s Jibunjishin^ o [s hanako\ ga [yp tj hihanshita ]]] (koto) self ACC Hanako NOM criticized fact 'Hanako criticized herself.' (Saito 1992: 76) Notice that Agrs is present in the above example. Since jibunjishin 'self is A'-scrambled, we expect that the presence of Agrs is not required following Miyagawa's proposal. However, when Agrs is absent, the sentence becomes much less acceptable, as shown in 28. (28) ? ? [ s Jibunjishin^ o [s hanako^ no [ V P tj hihanshita ]]] riyuu self ACC Hanako GEN criticized reason 'the reason Hanako criticized herself Furthermore, the acceptability of 28 does not improve much when the A'-moved direct object receives a heavy focus accent. Therefore, A'-scrambling is also subject to the presence of Agrs. A further problem with Miyagawa's analysis is that it is not clear as to whether an A'-moved constituent should receive a heavy focus accent. The example of A'-scrambling shown in 27 sounds perfect without the heavy focus stress on the A'-moved direct object NP. The sentence sounds also fine with the heavy focus stress on the subject NP. Furthermore, it is not necessary to 20Example 26a is my addition. 20 prepose a constituent with a focus accent. In the following example, 29b and 29c are the answers 01 to the question 29a. Direct objects in the answers are shown in bold face to indicate they are receiving the focus accent. (29) a. Jon wa donna shurui no niku o tabemasu ka? John TOP what kind of meat ACC eat Q 'What kind of meat does John eat?' b. Toriniku ox jon wa tj tabemasu. chicken ACC John TOP eat 'John eats chicken.' c. Jon wa toriniku o tabemasu. John TOP chicken ACC eat 'John eats chicken.' Since A-scrambling is not required, 29b must be A'-scrambling. The direct object with the focus accent may be preposed as in 29b. However, the sentence sounds perfect without preposing the 00 direct object as in 29c. Therefore, A'-scrambling cannot be explained by focus alone. 2.1.3 Summary Section 2.1 has reviewed syntactic analyses of OSV word order in scrambling. Section 2.1.1 has reviewed analyses of how scrambled constituents appear in the sentence-initial position. It also reviewed analyses of scrambling to two different types of positions; A- and A'-positions. I proposed that this thesis will follow the movement analyses (e.g. Saito and Hoji, 1983; Saito, 2 1 I employ the following diagnostic of focus, question/answer pair, from Rochemont (1998) in 29. In a well-formed information question/answer pair, the focus is the constituent in the answer that corresponds to the constituent that is wh-questioned in the question. (Rochemont 1998: 337) 2 2 The sentence sounds perfect as well when a wa-marked direct object that indicates a contrastive focus follows the subject as shown below: Jon wa toriniku wa tabemasu. John TOP chicken TOP eat 'John eats C H I C K E N (but he does not eat other kinds of meat).' 21 1985, 1992) and use the term "preposed" to indicate that the direct object has been moved to the sentence-initial position. Section 2.1.2 has reviewed an analysis of why some constituents are scrambled, and I have demonstrated counter-examples to the proposal that scrambling is motivated by agreement feature or focus. As we have seen, syntactic analyses explicate the mechanisms of scrambling. However, since these analyses are based on sentence grammar, they are unable to demonstrate the functions of scrambling beyond the sentence level. The following Section 2.2 reviews pragmatic analyses that attempt to explain the functions of preposing in discourse. It should be noted that I will not distinguish scrambling to A-position and A'-position, since these two positions cannot be distinguished in sentences unless the preposed constituent is an anaphor jibunjishin 'self or otagai 'each other' or an antecedent preceding a coindexed anaphor. 2.2 Pragmatic Analyses of Word Order Variations It is important to understand the characteristics of preposed constituents in order to understand the function of preposing. This section reviews two approaches to the analyses of preposing and related issues. Section 2.2.1 reviews an analysis that focuses on some characteristics of preposed constituents. Section 2.2.2 reviews analyses that focus on the information the preposed constituents convey, following a review of taxonomies employed to describe the information status of constituents proposed in previous studies. Section 2.2.3 summarizes the pragmatic analyses reviewed in Section 2.2. 2.2.1 Characteristics of Preposed Constituents Saeki (1960, 1975) examines the relative order of constituents in the written Japanese language and observes the following five tendencies in their order. 2 3 Saeki (1960) collected his samples from the following four stories: Kiiroifune by Murou Saisei, Komatta hibi by Yagi Yoshinori (first two episodes of four), Tokyo no tsuchi by Enchi Fumiko, and Shokujo by Takeuchi Kazuko. A l l appear in the collection Soosaku daihyoo sensyuu Vol . 22, edited by Nihon bungeika kyookai (1958). 22 (30) Word order tendencies in Japanese based on basic meanings and functions of constituents (Saeki I960):24 a. Locatives, such as constituents marked by particles ni 'to,' de 'at,' kara 'from,' and o 'on,' precede other constituents; b. constituents indicating time precede locatives; c. actors, constituents marked by nominative case marker ga, precede other constituents except for locatives; d. datives (or indirect objects) precede accusatives (or direct objects); e. locatives marked by the particle kara 'from' precede locatives marked by ni or e 'to.' Summarizing the tendencies described in 30, we can obtain the following ordering of major constituents in Japanese. (31) time --> locative —> nominative — > dative --> accusative — > verb Saeki claims that the word order in 31 describes* the neutral or unmarked order of the Japanese language because the tendencies in 30 are observed when NPs do not have a special meaning or function aside from their basic ones, or when certain characteristics of sentence constituents do not influence word order. In order to obtain data with such NPs, he excluded NPs that a) consist of different numbers of bunsetsu units;25 b) contain demonstratives;26 c) are followed by kakari-97 joshi; d) are closely related to one another in their meanings; and e) convey idiomatic meanings 2 4 The translations of citations from sources written in Japanese are mine unless noted. 2 5 T h e "bunsetsu unit" is equivalent to syntactic constituents such as Determiner phrases (DPs) and verbs. Hashimoto (1948: 6) defines bunsetsu as "the unit we first obtain by dividing sentences" and "the smallest phrase we obtain by dividing sentences into as many units as possible." See Appendix B for further description of the bunsetsu unit. 2 6 See Appendix B for the definition of "demonstrative". Also, Section 2.3.2 explains what this thesis refers to by direct objects containing demonstratives. I1 Kakari-joshi comprise particles that affect the entire predication, and which figure prominently in the classical language of the Heian period (Shibatani 1990). Kakari-joshi are particles such as wa (topic or contrastive), mo 'also,' koso (emphatic), sae 'even,' demo 'or something,' hoka 'other than,' and shika 'only.' Some theories treat only wa and mo as kakari-joshi. 23 with verbs. He examined the word order of the NPs which fall under descriptions (a-e) separately and observed the following tendencies.28 (32) Word order tendencies in Japanese based on the characteristics of constituents (Saeki 1960): a. Long NPs tend to precede short NPs; b. NPs containing demonstratives tend to precede the ones that do not; c. when an NPi is closely related in its meaning to an NP2 and the NPi modifies the NP2, the NPi precedes the NP2; d. an NP immediately precedes a verb when the two constituents form an idiomatic expression. Saeki provides sample sentences from his data for each environment to demonstrate the tendencies in 32. The following reviews each of the samples from Saeki. In the environment in which the lengths of constituents in a sentence are different, Saeki observes that long NPs tend to precede short NPs. 2 9 Example 33, from Saeki, shows a long object preceding a shorter subject:30 2 8 Saeki (1960, 1975) does not discuss the word order tendency in environment (c), in which constituents are followed by kakarl-joshl. ' „ 2 9Movement of lengthy NPs occurs in English as well, and it is called "Heavy NP Shift." In English, a lengthy (or heavy) NP is postposed to the sentence-final position as shown in example below: Kelly bought for Sam last week a brand new computer. (Rochemont 1998: 339) The ditransitive sentence normally appears in subject -> verb -> dative phrase -> accusative phrase order. The underlined direct object 'a brand new computer' is postposed to the sentence-final position due to Heavy NP Shift in the above example. It should be noted that a postposed NP in English Heavy NP Shift conveys a focus, and thus Heavy NP Shift identifies a structural focus. 3 0 The NPs in question are indicated by underlines in the examples. 24 (33) Tsuya no ban ni atsumattekita kinio no shufutachi no wake GEN evening on gathered neighbourhood GEN housewives GEN nakade kuchi no ooi no de yoku izakoza no moto o among mouth GEN many NMR because of often argument GEN cause ACC tsukuru hvoogushi no saikun ga, "hontooni kawaisooni. . . . Onaji create paper hanger GEN wife NOM really unfortunate same mononara obaasan to kawareba ne. " to itta no o NMR-ifitwere grandmother with if-take-the-place-of SFP QT said NMR ACC nishime no sara o kae-ni-kita obaasan ga senakaawase ni nishime GEN plate ACC came-to-change grandmother NOM back to back LOG ite tashikani kiiteita to • yuu. was-there surely heard QT say 'It is said that the grandmother, who came to change the nishime plate and was behind the others, heard the paper hanger's wife who, among the housewives of the neighbourhood who gathered on the evening of the wake, was often the cause of arguments because of her talkativeness, saying, "It is really unfortunate.... It would have been better if the grandmother had taken his place (and died).'" (Tokyo no tsuchi by Enchi Fumiko: 353) (cited in Saeki 1960: 61) Based on the neutral order described in 31, we expect the underlined direct object phrase, tsuya no ban ni atsumattekita kinjo no shufutachi no nakade kuchi no ooi no de yoku izakoza no moto o tsukuru hyoogushi no saikun ga, "hontooni kawaisooni. . . . Onaji mononara obaasan to kawareba ne." to itta no o 'the paper hanger's wife who, among the housewives of the neighbourhood who gathered on the evening of the wake, was often the cause of arguments because of her talkativeness, saying, "it is really unfortunate. . . . It would have been better if the grandmother had taken his place (and died).'" to follow the subject phrase, nishime no sara o kae-ni-kita obaasan ga 'the grandmother, who came to change the nishime plate NOM.' However, the direct object precedes the subject in 33 above. Saeki explains that such word order variation occurs in order to avoid ambiguity; a clause becomes less intelligible when a long NP, particularly one containing several clauses, intervenes between another NP and a verb within the 25 same clause/ Saeki demonstrates this in the following example, in which the order of the subject and the direct object in example 33 is reversed: (34) Nishime no sara o kae-ni-kita obaasan ga, tsuya no nishime GEN plate ACC came-to-change grandmother NOM wake GEN ban ni atsumattekita kinio no shufutachi no nakade evening on gathered neighbourhood GEN housewives GEN among kuchi no ooi no de yoku izakoza no moto o tsukuru mouth NMR many NMR because often argument NMR cause ACC create hyoogushi no saikun ga. "hontooni kawaisooni. . . . Onaii paper hanger GEN wife NOM really unfortunate same mononara obaasan to kawareba ne. " to itta no o . NMR-ifitwere grandmother with if-take the place-of SFP QT said NMR ACC senakaawase ni ite tashikani kiiteita to yuu. back to back LOC was there surely heard QT say (Saeki 1960: 61-62) Saeki compares 33 and 34 and argues that in 34, it is difficult to associate the verb ite 'was there' and kiiteita 'heard' with their agent obaasan 'grandmother' because the lengthy direct object intervenes between the verbs and their agent. Notice that the direct object in 34 contains the clause tsuya no ban ni atsumattekita kinio no shufutachi no nakade kuchi no ooi no de yoku izakoza no moto o tsukuru hyoogushi no saikun ga, "hontooni kawaisooni. . . . Onaji mononara obaasan to kawareba ne." to itta 'the paper hanger's wife who, among the housewives of the neighbourhood who gathered on the evening of the wake, was often the cause of arguments because of her talkativeness, said, "It is really unfortunate. . . . It would have been better if the grandmother had taken his place (and died).'" The subject of this intervening clause, hyoogushi no saikun 'the paper hanger's wife,' immediately follows the subject obaasan 'grandmother' in 34; therefore, it is difficult to associate obaasan 'grandmother' and the verbs ite 'was there' and kiiteita 'heard.' As a result, sentence 34 is difficult to follow. Saeki argues that in order to avoid 3 1Frazier and Rayner (1988) measure the time taken to read English sentences containing constituents appearing in different orders, such as 'That both of the Siamese twins survived the operation is remarkable' and 'It is remarkable that both of the Siamese twins survived the operation,' to examine the influence of orders of constituents and processing. 26 producing the less intelligible sentence 34, the direct object is preposed to the sentence-initial position as in 33. Kuno (1973) also argues that lengthy constituents are preposed in order to avoid ambiguity. He explains that difficulty following a sentence containing a clause intervening between the subject and the verb of the main clause, such as the one observed in 34, is due to the structure of noun modification in Japanese, where clauses precede the nouns they modify, as shown in 35. (35) Jon ga katteiru neko ga koroshita nezumi ga tabeta John NOM keep cat NOM killed rat NOM ate chiizu wa kusatteita. cheese TOP rotten. (Literal word order) 'John keeps cat killed rat ate cheese was rotten.' 'The cheese that the rat that the cat that John keeps killed ate was rotten.' (Kuno 1973: 7) The Japanese sentence in 35 is perfectly grammatical and intelligible while its English translation is not. As opposed to the noun modification structure in Japanese, NPs precede noun-modifying clauses in English as shown in 36. (36) John owned a cat that killed a rat that ate cheese that was rotten. (Kuno 1973: 8) Kuno argues that English sentences such as 36 are intelligible even with a lengthy direct object since the modifying clauses follow the modified direct object and thus do not intervene between the verb and the modified direct object as in the translation of 35. Below is another example from Kuno: (37) John read the letter that Mary had written to the boy that Jane loved. (Kuno 1973: 10) The direct object 'the letter' is modified by a clause 'that Mary had written to the boy that Jane loved' which contains two verbs and is thus long, but completely intelligible. On the other hand, its Japanese counterpart 38a is not intelligible, but improves when the direct object is preposed as in 38b:32 32Example 38b becomes more intelligible if the dative clause Jeen ga aishiteiru shoonen ni 'to the boy Jane loves' appears in the sentence-initial position followed by the direct object and the subject as shown below: 27 (38) a. Vohn ga [ marii ga jeen ga aishiteiru shoonen ni kaita John NOM Mary NOM Jane NOM love boy DAT wrote tegami o ] yonda. letter ACC read 'John read the letter that Mary wrote to the boy that Jane loved.' (Kuno 1973: 10) b. [Marii ga jeen ga aishiteiru shoonen ni kaita tegami o ] Mary NOM Jane NOM love boy DAT wrote letter ACC jon ga yonda John NOM read 'John read the letter that Mary wrote to the boy that Jane loved.' (Kuno 1973: 10) As demonstrated in Saeki's and Kuno's analyses, long constituents, particularly those containing several clauses, are preposed in order to produce intelligible sentences. In the environment in which NPs contain demonstratives, Saeki observes that such NPs tend to precede the ones that do not. Example 39 from Saeki illustrates a sentence in which a direct object containing a demonstrative precedes a subject. (39) Sore o iguchi ga uketa no da to wa. . . it ACC Iguchi NOM had NMR CPL COMP TOP 'that Iguchi had it (=the operation)...' (Shokujo by Takeuchi Kazuko: 315-316) (cited in Saeki-1960: 62) Subjects normally precede direct objects; however, the direct object phrase sore o 'it A C C precedes the subject phrase iguchi ga 'Iguchi NOM' in 39. Saeki explains that constituents containing demonstratives are preposed in order to make sentences less ambiguous. The direct object sore 'it' refers to an operation mentioned in the preceding discourse and is preposed in order to make this reference clearer. In the environment in which two constituents are closely related in meaning to one another, Saeki observes that an NPi closely related to an NP2 tends to precede the NP2. He provides the following example to demonstrate this tendency. Jeen ga aishiteiru shoonen ni marii ga kaita tegami o jon ga yonda. Jane N O M love boy D A T Mary N O M wrote letter A C C John N O M read. 'John read the letter that Mary wrote to the boy that Jane loved.' 28 (40) Kono konsetsu-teeneena choobun no tegami ni 'kotowari-ioo' o this cordially-polite long of letter DAT declining-letter ACC kaku yorimo write rather than 'Rather than writing a letter of refusal to this cordially polite long letter' {Komatta hibi by Yagi Yoshinori: 334) (cited in Saeki 1960: 62) Saeki explains that the direct object kotowari-ioo 'letter of refusal' is a letter replying to kono konsetsu-teeneena choobun no tegami 'this cordially polite long letter' in 40. In other words, the dative NP modifies the accusative NP because of the meanings they convey. Therefore, shifting the order of the dative NP and accusative NP, as shown in 41 below, results in an unnatural sentence. (41) ? 'Kotowari-ioo' o kono konsetsu-teeneena choobun no tegami ni declining-letter ACC this cordially-polite long of letter DAT kaku yorimo write rather than 'Rather than writing a letter of refusal to this cordially polite long letter' Saeki observes a similar tendency in example 42 below: (42) Kawa-nami ga maue ni hi o uketeiru kirakira o river-wave NOM right above LOC sun ACC getting ONM ACC 'the glitter of the waves of the river getting sun right above' (Kiiroifune by Murou Saisei: 386) (cited in Saeki 1960: 62) He explains that maue 'right above' is kawa-nami no maue 'right above the waves of the river' in 42; thus, kawa-nami 'the waves of the river' and maue 'right above' appear in the order shown in 42. As demonstrated in the comparison of examples 40 and 41, reversing the order of kawa-nami 'the waves of the river' and maue ni 'right above' results in an unnatural sentence as shown below: (43) ? Maue ni kawa-nami ga hi o uketeiru kirakira o right above LOC river-wave NOM sun ACC getting ONM ACC 'the glitter of the waves of the river getting sun right above' 29 Following the neutral word order described in 31, we expect the locative phrase maue ni 'right above' to precede the subject phrase kawa-nami ga 'river-waves NOM' as in 43. Saeki argues that the neutral order of locative -> nominative appears shifted as in 42 in order to clearly present the relationship of the meanings of these constituents, namely that it is kawa-nami no maue 'right above the waves of the river.' Notice that Saeki observes such a close relationship between a dative NP and a direct object as in 40, and a locative NP and a subject as in 42. The subject and the direct object do not exhibit such a relationship. In the environment in which an NP and a verb together form an idiom, Saeki observes that the NP immediately precedes the verb. He demonstrates this in example 44 below: (44) a. Sumida-gawa e mi o nagete Sumida-river LOC body ACC throw 'committing suicide by throwing oneself into the Sumida river' (Tokyo no tsuchi by Enchi Fumiko: 353) (cited in Saeki 1960: 63) b. Sakuya no koto o ne ni motta oshii wa last night GEN event ACC root LOC had (resent) Oshii TOP 'Oshii, who resents what happened the night before' (Kiiroifune by Murou Saisei: 381) (cited in Saeki 1960: 63) The underlined parts, mi o nagete 'committing suicide by throwing oneself in 44a and ne ni motta 'resent' in 44b, are a combination of an NP and a verb, and together they form idiomatic expressions. Saeki claims that when other NPs intervene between NPs and verbs with which they form idioms, the sentences become unnatural. In the below example the order of the constituents in 44a and 44b are changed, and an NP separates the NP and the verb forming the idiomatic expression.' (45) a. ?mt o Sumida-gawa e nagete body ACC Sumida-river LOC throw 'committing suicide by throwing oneself into the Sumida river' b. -ne ni sakuya no koto o motta oshii wa root DAT last night GEN event ACC have Oshii TOP 'Oshii, who resents what happened the night before' 30 The idiomatic meanings 'committing suicide by throwing oneself and 'resent' are lost in 45a and 45b respectively due to NPs intervening between NP-verb pairs that form idiomatic expressions. Therefore, we would expect a direct object forming an idiomatic expression together with a verb, such as mi o 'body A C C of mi o nagete 'committing suicide by throwing oneself in 44a, to tend not to appear in the sentence-initial position. As previously mentioned, Saeki analyzes the word order tendencies of written Japanese with two sets of data. One of the merits of his data analysis method is that he is able to observe neutral word order by removing NPs with special structural characteristics from the first set and treating them separately; namely those NPs which a) contain different number of bunsetsu units; b) contain demonstratives; c) are accompanied by kakari-joshi; d) are closely related to one another in their meanings; and e) convey idiomatic meanings with verbs. Another merit is that by isolating these NPs he can analyze their influence alone on word order tendencies. On the other hand, since NPs with these characteristics are analyzed separately, he does not obtain the frequency of word order variation due to. those particular characteristics of NPs. 3 3 Therefore, his analysis lacks statistical evidence to support his claims. This section has reviewed Saeki's (1960, 1975) analyses, which focus on the influences of some characteristics on word order variations. We have seen that neutral word order in Japanese is as follows: (46) time — > locative — > nominative --> dative — > accusative --> verb (=31) We have also seen that some characteristics of constituents tend to shift the neutral word order in 46. The characteristics Saeki observes are repeated in 47 below: (47) Word order tendencies in Japanese based on the characteristics of constituents (Saeki 1960): (=32) a. Long NPs tend to precede short NPs; 3 3 Saeki (1960) observes twenty-two clauses with SO order but does not find clauses with OS order in the set of data that excludes NPs which consist of different numbers of bunsetsu units; contain demonstratives; are followed by kakari-joshi; are closely related to one another in their meanings; and convey idiomatic meanings with verbs. He does not specify how many clauses with SO and OS orders are found in clauses containing the above NPs. 31 (47) b. NPs containing demonstratives tend to precede the ones that do not; c. when an NPi is closely related in its meaning to an NP2 and the NPi modifies the NP2, the NPi precedes the NP2; and d. an NP immediately precedes a verb when the two constituents form an idiomatic expression. We have seen that the above characteristics of constituents shift the neutral word order in order to make sentences easy to follow. Based on the review of Saeki's above claims, we expect the OSV word order variation to have the following tendencies: (48) a. A long direct object tends to precede a short subject; b. a direct object containing a demonstrative tends to precede a subject that does not; c. a direct object forming an idiomatic expression together with a verb tends not to be preposed. Saeki's analysis provides some explanations for OSV word order at the level of sentences. The next section reviews analyses of preposing that look at the preposing of constituents in context. 2.2.2 Information Status of Preposed Constituents It has been proposed that the types of information the constituents in a sentence convey determine the order they appear in that sentence. For example, some researchers have proposed that a sentence is organized in such a way that constituents conveying old information precede those conveying new information (e.g. Firbas, 1964; Halliday, 1967; Kuno, 1978b; Givon, 1983, 1988; Gundel, 1988; Cowan, 1995), or those conveying new or important information precede others (Mithun, 1986; Givon, 1988; Gundel, 1988). Preposing is also analyzed with respect to the types of information preposed constituents convey. For example, Kuno (1973) proposes that preposed constituents are emphasized in Japanese, and Ward (1988) and Birner and Ward (1998) propose that preposed constituents are anaphorically linked to the preceding discourse in English. The first part of this section reviews terms and definitions employed in analyzing the information status of the preposed constituents. The second part reviews an analysis of preposing in Japanese that also focuses on the information the preposed constituents convey. Discourse entities were classified into two types in the 1960s and the 1970s, namely, 32 entities conveying new and old information. The two types are described employing various terms, and often interpreted in various ways. For example, Firbas (1964) describes new and old information employing the terms "rheme" and "theme" respectively. He introduces 'communicative dynamism' (CD), which shows the extent to which the constituents in a sentence contribute to the development of the communication. He classifies constituents carrying a higher degree of CD, which convey new or unknown information, as "rheme," and those carrying a lower degree of CD, which convey known information, as "theme." Halliday (1967) uses the terms "new" and "given," and defines them as an entity with an intonationally marked or unmarked focus, and the complement of that focus, respectively. Some researchers point out that what is "new" or "old" is classified based on the speaker's assumption on the hearer. For example, Haviland and Clark (1974) and Clark and Haviland (1977) define "new" information as "what the speaker thinks his audience does not already know," and "given" information as "what the speaker thinks his audience already knows" (Haviland and Clark 1974: 513). For Chafe (1976), consciousness of the hearer is a key for distinguishing "given" from "new" information. In his terms, a "given" item of information is "knowledge which the speaker assumes to be in the consciousness of the addressee at the time of the utterance," and "new" information is "what the speaker assumes he is introducing into the addressee's consciousness by what he says" (1976: 30). Kuno (1978c) defines "new" and "old" information with respect to preceding discourse, and explains that "an element in a sentence represents old, predictable information if it is recoverable from preceding context; if it is not recoverable, it represents new, unpredictable information" (1978: 282-283). The table below summarizes the terms and definitions employed in the above previous studies. 33 New information Old information Firbas (1964) Rheme: • new, unknown information Theme: • known information, • information that can be gathered from the verbal or situational context Halliday (1967) New: • intonationally marked or unmarked focus Given: • complement of marked focus Haviland and Clark (1974), Clark and Haviland (1977) New: • information the speaker thinks the hearer does not already know Given: • information the speaker thinks the hearer already knows Chafe (1976) New: • knowledge which the speaker assumes to be in the consciousness of the hearer at the time of the utterance Given: • what the speaker assumes he/she is introducing into the hearer's consciousness by what he/she says Kuno (1978c) New: • information that is not recoverable from the preceding context Old: • information that is recoverable from the preceding context Table 1. Summary of definitions of new and old information in previous studies There are some problems with these taxonomies. For example, Firbas (1964) recognizes that CD has degrees; however, the entities are divided into only two categories: "theme" and "rheme." Some classify entities as "given" with respect to the preceding discourse (Kuno 1978c), and some with respect to the hearer's knowledge (Haviland and Clark, 1974; Clark and Haviland, 1977; Chafe, 1976). Prince (1981) points out the problems in the dichotomies employed in previous studies such as those summarized in Table 1 and also discusses the problem by which the term "given" is used to refer to different notions. She argues that "given" is used to refer to predictability or recoverability (cf. Halliday, 1967; Kuno, 1978c), shared knowledge (cf. Haviland and Clark 1974), and saliency (cf. Chafe 1976). She further argues that all three notions are ultimately 34 related to "extralinguistic phenomena, in particular to what the speaker thinks is, should, or could appropriately be in the hearer's mind" (1981: 232). She proposes using the term "Assumed Familiarity" to classify information statuses instead of "given/new." Besides what is "given" ("Evoked" is Prince's term) and what is "new," she recognizes that the hearer has some assumption of or can draw some inference from what he or she knows, and she then introduces the third type of information, "Inferrable." She further classifies three types of "Assumed Familiarity" ("New," "Inferrable," and "Evoked"), into seven types as shown in Figure 1: Assumed Familiarity Brand-new Unused (Noncontaining) Containing (Textually) Situationally Inferrable Inferrable Evoked Evoked Brand-new Brand-new (Unanchored) Anchored (=Prince 1981: 237) Figure 1. Assumed familiarity A "Brand-new" NP is a newly introduced entity. The bold faced NP 'a bus' in 49 is an example from Prince of a "Brand-new" NP. 3 4 (49) "Brand-new (Unanchored)" I got on a bus yesterday and the driver was drunk. (Prince 1981: 233) The second type of "Brand-new" NP is the "Brand-new Anchored" NP, which represents a newly introduced entity linked by means of another NP, or "anchor," properly contained within it, to some other discourse entity. Thus, 'a guy I work with' in 50 is a "Brand-new Anchored" NP linked by the NP T.' 3 4The use of bold face in the examples 49 to 54 is Prince's (1981). 35 (50) T3rand-new Anchored' A guy I work with says he knows your sister. (Prince 1981: 233) An NP.is "Unused" when the hearer may be assumed to have a corresponding entity in his/her own model and simply has to place it in (or copy it into) the discourse-model. The NP 'Noam Chomsky' below is an example of an "Unused" NP. (51) "Unused" An "Inferrable" NP is an entity in which the speaker assumes the hearer can infer its referent, via logical reasoning, from discourse entities already evoked or from other inferrables. The NP 'the driver' in 49, repeated in 52, is an "Inferrable," in which the hearer can infer its referent from the already introduced NP 'a bus.' (52) "(Noncontaining) Inferrable" I got on a bus yesterday and the driver was drunk. (Prince 1981: 233) A "Containing Inferrable" is an entity where what is inferenced is properly contained within the inferrable NP itself. The NP 'one of these eggs' in 53 is an example of a "Containing Inferrable." (53) "Containing Inferrable" Hey, one of these eggs is broken! (Prince 1981: 233) There are two types of "Evoked" NPs: "(Textually) Evoked" and "Situationally Evoked." A "(Textually) Evoked" NP is an entity the hearer evoked previously, on textual grounds, by following instructions from the speaker. The NP 'he' in 54a is "(Textually) Evoked" by the NP 'a guy I work with' which was evoked earlier. "Situationally Evoked" NPs represent discourse participants and salient features of the extratextual context, which includes the text itself. Thus, the NP 'you' in 54b is "Situationally Evoked." (54) a. "(Textually) Evoked" Noam Chomsky went to Penn. (Prince 1981: 233) A guy I work with says he knows your sister. (Prince 1981: 233) 36 (54) b. "Situationally Evoked" Pardon, would you have change for a quarter? (Prince 1981: 233) Prince (1992) further shows that the information status in regard to both the discourse and the hearer are closely related and thus categorizes discourse entities accordingly. Shown below is her 1992 categorization: Discourse-new Discourse-old Hearer-new Brand-new (Unanchored) Brand-new Anchored Hearer-old Unused Table 2. Information status of NPs In the above table Prince shows that what is new to the hearer must be new to the discourse, and that what is old to the discourse must be old to the hearer as well; however, what is old to the hearer is not necessarily old to the discourse, and what is new to the discourse is not necessarily new to the hearer. The following example from Prince demonstrates the difference between the "Hearer-old/Hearer-new" and "Discourse-old/Discourse-new" status. In each example, the same speaker utters parts "a" and "b" consecutively. The NPs in the discussion are indicated with italics.35 (55) a. I'm waiting for it to be noon so I can call someone in California. b. I figure she'll be up by 9, her time. (Prince 1992: 309) (56) a. I'm waiting for it to be noon so I can call Sandy Thompson. b. I figure Sandy/she'll be up by 9, her time. (Prince 1992: 309) The NPs 'she' in 55b and 'Sandy/she' in 56b, which are "Textually Evoked," are both old with 3 5 The italicization is from Prince's (1992) original. 37 (Textually) Evoked Situationally Evoked (=Prince 1992: 309) respect to both the discourse and the hearer; the referents of these NPs were already evoked in their preceding discourses (55a and 56a). The NPs 'someone in California' in 55a and 'Sandy Thompson' in 56a were both introduced into the discourse for the first time at the time of the utterances; therefore, they are both "Discourse-new." However, the former is "Hearer-new" and the latter is "Hearer-old." The speaker of utterance 55a uses the NP 'someone in California' to describe whom she or he is calling because she or he assumes that the hearer does not know who this person is. On the other hand, the speaker of the utterance 56a presumes that the hearer already knows who Sandy Thompson is, and therefore uses the NP 'Sandy Thompson.' Prince pays attention to some characteristics of the "Inferrable" entities and thus categorizes them separately from "Discourse-new/old" and "Hearer-new/old" status. "Inferrables" are introduced to the discourse for the first time at the point when they are uttered; therefore, they are "technically Hearer-new and Discourse-new" (1992: 307). However, they are different from "Hearer-new" and "Discourse-new" entities ("Brand-new" entities) in the way that they are inferrable from "Discourse-old" trigger entities. Chafe (1987) introduces a third type of information status besides "given" and "new." He claims that "our minds contain very large amounts of knowledge or information, and that only a very small amount of this information can be focused on, or be 'active' at any one time" (1987: 22), and characterizes "given" and "new" information in terms of the "activation" status of information. In his terms, "already active" and "previously inactive" information correspond to "given" and "new" information respectively. The third type he proposes is "semi-active" (or "accessible") information, which is in person's peripheral consciousness. An entity which conveys "semi-active" information is one in which "a person has a background awareness, but which is not being directly focused on" (1987: 25). Previously semi-active information refers to information that is in a person's peripheral consciousness, and is thus intermediate between "given" and "new" information. Dryer (1996) further characterizes Chafe's (1987) notion "activation" as being a continuum rather than as having discrete stages. He distinguishes two phases of "activated" information: 38 "focus of attention" and "activated but not focus of attention," which were both "active" in Chafe. Figure 2 is a summary of the continuum of activation from Dryer: focus of attention > activated but not focus of attention > recently activated but now semi-deactivated > nonactivated or accessible to activation (=Dryer 1996: 482) Figure 2. Continuum of activation Dryer's continuum has four different phases: focus of attention, activated but not focus of attention, recently activated but now semi-deactivated, and nonactivated. Dryer claims that nonfocus is associated with activation, not presupposition (cf. Chomsky, 1972; Jackendoff, 1972). Nonfocused parts of an utterance correspond to activated information and focused parts to information that is nonactivated or less activated in the immediately preceding context. Although Chafe's (1987) and Dryer's (1996) notion of "activation" describes the statuses of information in the hearer's consciousness more accurately than the "given/new" notion, it is vague in that it does not distinguish what is "Brand-new" and what is "Unused" as the taxonomy proposed in Prince (1981). It also does not describe the information status of constituents with respect to the discourse as does Prince's. On the other hand, Prince's taxonomy does not distinguish the. different statuses of "Evoked" information such as "active" and "semi-active" in Chafe's taxonomy. Lambrecht (1994) attempts to incorporate Prince's (1981) seven types of "Assumed Familiarity" and Chafe's (1987) "activation" statuses of information by introducing the new notion "identifiability." He argues that information is either "identifiable" or "unidentifiable," and that "identifiable" information is further classified into three stages of activation. Figure 3 is a summary of the systems of identifiability and activation from Lambrecht: 39 IDENTMABILITY unidentifiable unanchored anchored identifiable ACTIVATION inactive active textually situationally inferentially (=Lambrecht 1994: 109) Figure 3. Identifiability of information By incorporating Prince's and Chafe's taxonomy, Lambrecht describes detailed statuses of information as shown in Figure 3. However, the degree of activity of the information in the hearer's consciousness, which is important to the notion of "activation," is lost in his taxonomy. There are some proposals that preposing in Japanese is due to the information the preposed constituents convey. Kuno (1973) and Saeki (1989) propose that constituents can appear in the sentence-initial position when they are emphasized in Japanese; however, they do not provide examples of data supporting their proposals. Masunaga (1983) proposes that the information the preposed constituents convey must have been already mentioned in the preceding discourse in Japanese. She illustrates this proposal in the following example:36 (57) a. w?Hitori no onnanohito o taroo ga nagutta. one GEN woman ACC Taro NOM hit 'Taro hit a woman.' b. Sono onnanohito o taroo ga nagutta. that woman ACC Taro NOM hit 'Taro hit that woman.' (Masunaga 1983: 455) (Masunaga 1983: 456) 36Translations of example Japanese sentences from Masunaga (1983) are her original. 40 She argues that 57a is awkward because the preposed direct object hitori no onnanohito 'a woman' is indefinite, and thus has no referent in the preceding discourse. On the other hand, the sentence improves when we change the preposed constituent from indefinite to definite as in 57b. The demonstrative sono 'that' indicates that the direct object in 57b onnanohito 'woman' refers to a woman who has already been introduced in the preceding discourse. Masunaga claims that 58 below cannot be used unless the preposed direct object Hanako 'Hanako' is mentioned in the preceding discourse. (58) Hanako o taroo ga nagutta. Hanako ACC Taro NOM hit 'Taro hit Hanako.' (Masunaga 1983: 456) Masunaga also claims that preposed constituents do not have to be identical to the constituents mentioned in the preceding discourse; thus, 59 is acceptable as well. (59) A (detective): Hanako wa kaishain-fuu no otoko to atte-inakatta Hanako TOP office worker-like GEN man with meet-didn't ka ne Q SFP 'Didn't Hanako meet a man who appears to be an office worker?' B (witness): Soo desu ne. . . Soo ieba dareka to hanako wa that CPL SFP that if I say somebody with Hanako TOP nagai aida hanashiteita yoo desu ga. . . long time talking seem CPL but 'Well... Come to think of it, Hanako seemed to be talking with somebody for a long time, but. . . ' (Masunaga 1983: 457) In the above example, dareka to 'with somebody' is preposed in B's utterance, but dareka 'someone' is not mentioned in the preceding discourse (A's utterance). Masunaga explains that the NP dareka 'somebody' is already mentioned in a different expression kaishain-fuu no otoko 'a man who appears to be an office worker' in A's utterance; therefore, preposing dareka to 'with somebody' is acceptable. B's utterance is also acceptable when the preposed dareka 'someone' does not refer to kaishain-fuu no otoko 'a man who appears to be an office worker' as shown in 41 60 below:J1 (60) B (witness): Soo desu ne. . . Soo ieba dareka to hanako wa that CPL SFP that if I say somebody with Hanako TOP nagai aida hanashiteita yoo desu ga. . . long time talking seem CPL but Kaishain no yoo dewaarimasendeshtia. office worker of seem CPL-negative-past 'Well... Come to think of it, Hanako seemed to be talking with somebody for a long time, but... He did not seem to be an office worker.' The last sentence of B's utterance Kaishain no yoo dewaarimasendeshita 'He did not seem to be an office worker' in 60 excludes kaishain-fuu no otoko 'a man appears to be an office worker' from the description of the person Hanako met. Therefore, dareka 'someone' in 66 does not refer to kaishain-fuu no otoko 'a man who appears to be an office worker,' thus the preposed dareka 'someone' is not mentioned the preceding discourse. Hence, it will be more appropriate to propose that the preposing of dareka to 'with someone' is acceptable because the existence of 'someone' is implied in the preceding discourse. Masunaga further argues that NPs which are not mentioned in the preceding discourse but appear in a nonlinguistic context can be preposed as well. Example 61 is from Masunaga: (61) Ano eega o taroo wa jyuugo-kai mo mita-n-da-tte. that movie ACC Taro TOP fifteen-NC EMPH saw-NMR-CPL-heard T heard that Taro saw that movie fifteen times.' (Masunaga 1983: 457) She explains that sentence 61 can be uttered looking at a poster of a movie on a wall far away from both the speaker and the hearer. The speaker opens the conversation by uttering 61, thus the direct object ano eega 'that movie' is not mentioned in the previous discourse. The utterance 61 is acceptable because the preposed direct object ano eega 'that movie' has an antecedent in the nonlinguistic context, a poster. Therefore, she claims that NPs which appear in a nonlinguistic context can be preposed as well as NPs mentioned in the preceding discourse. 3 7Collier-Sanuki (personal communication) brought this point to my attention. 42 Based on the observations of such characteristics of preposed constituents, Masunaga proposes that such preposed constituents perform a "bridging function" which provides "a 'bridge' between the preceding discourse and the rest of the present sentence" (1983: 456). Therefore, preposed constituents are the "bridges" which connect the preceding discourse with the sentences in which they occur. Masunaga also claims that by preposing NPs such as objects, the topic of the sentence in question is established. In other words, a preposed constituent can establish the topic without being marked by the thematic particle wa. She further claims that topicalization and establishing topics by preposing are subject to different conditions. However, she does not discuss what conditions apply to topicalization. As we have seen, Masunaga describes the information preposed constituents convey as the information "mentioned in the preceding discourse" (1983: 456). This expression is ambiguous because it refers to broad types of information including what is implied. It would also involve "given" information in terms of Halliday (1967), Haviland and Clark (1974), Clark and Haviland (1977), and Chafe (1976), and "old" information as in Kuno's (1978c) term. Furthermore, preposed constituents are not always "mentioned" as shown in example 67: the preposed direct object ano eega 'that movie' is not mentioned in the preceding discourse, but it is "Situationally Evoked" in Prince's term. Therefore, Masunaga's description of the information the preposed constituents convey, "mentioned in the preceding discourse," is not accurate, and thus needs to be modified so that the preposed constituents are linked to the constituents in the preceding discourse or entities in the nonlinguistic context. This section has reviewed the terms and definitions employed to describe the information status of the constituents in previous studies. We saw that such status of entities in discourse were defined employing many different terms. Furthermore, we saw that even the same terms, "given" and "new," refer to different notions when used by different scholars. This section also reviewed Masunaga's (1983) analysis, which examines the relationship between the preposed constituents and the preceding discourse. We have seen that some preposed constituents are already mentioned in the preceding discourse. We have also seen that there were several types of "previously 43 mentioned" preposed constituents such as one whose antecedent is mentioned in the previous discourse, and one whose antecedent exists in a nonlinguistic context. Therefore, it is important to provide a clear definition when describing the information status of entities in discourse. 2.2.3 Summary This section has reviewed some previous studies focused on the characteristics of preposed constituents and the information status of discourse entities. Section 2.2.1 reviewed Saeki's (1960, 1975) analyses of word order in written Japanese. The findings in his study show that some characteristics of constituents, such as length and the containing of demonstratives, have some influence on basic word order. They also show that the preposing of some constituents occurs in order not to produce unintelligible sentences. Section 2.2.2 reviewed the literature on information status of discourse entities. It shows that there are many different terms which classify information status, and different notions are described using the same terms. Section 2.2.2 also reviewed Masunaga's (1983) analysis, which also focuses on the characteristics of preposed constituents. It shows that the relationship between the preposed constituents and the preceding discourse or entities in nonlinguistic context influences word order. It also shows that it is important to define terms clearly when describing the information status of discourse entities. Based on these reviews of previous studies, the following section proposes the hypotheses and taxonomical framework of this thesis. 2.3 Hypotheses and Taxonomical Framework of this Thesis As we have seen in the previous chapter, understanding the characteristics of the constituents of sentences is an essential step towards understanding the function of word order variation. In order to understand the function of OSV word order, this thesis examines the relationship between the characteristics of direct objects and the preposing of direct objects. Based on the above foundation, this chapter presents the hypotheses that this thesis examines in Section 2.3.1, and then the terms and the taxonomy this thesis employs in Section 2.3.2. Section 2.3.3 44 summarizes the hypotheses of this thesis. 2.3.1 Hypotheses of this Thesis In Section 2.2 we saw the following propositions which characterize word order variation due to preposing in Japanese: (62) a. Long NPs tend to precede short NPs (Saeki 1960, 1975); b. NPs containing demonstratives tend to precede the ones that do not (Saeki 1960, 1975); c. an NP forming an idiomatic expression together with a verb tends not to be preposed (Saeki 1960, 1975); d. emphasized NPs are preposed (Kuno, 1973; Saeki, 1989); and e. preposed NPs must have been linked either to entities in the preceding discourse or those in the nonlinguistic context (Masunaga 1983). This thesis attempts to verify that the above claims, which refer to preposing in Japanese in general, can also be applied to OSV word order, in which the direct object is preposed to the sentence-initial position. Based on the claims listed in 62, this thesis hypothesizes that the following characteristics are observed in the direct objects in the OSV word order: (63) The direct objects in OSV clauses a. tend to be lengthy compared to those in SOV clauses; b. tend to contain demonstratives unlike those in SOV clauses; c. tend not to form an idiomatic expression together with a verb; d. tend to be an emphasized NP; and e. tend to be closely related to entities in the preceding discourse or entities in the nonlinguistic context. As pointed out in Section 2.2.2, Masunaga (1983) uses the expression "mentioned in the preceding discourse" as broadly applying to entities such as those visible in the extralinguistic context and implied in the preceding discourse. Therefore, this thesis examines the relationship between preposed NPs and their preceding discourses as described in 63e. In order to examine these five 45 hypotheses, this thesis will analyze data collected from written discourses.3 The following section explains the terms and taxonomy this thesis employs in examining the above hypotheses. 2.3.2 Terms and Taxonomy Employed in this Thesis This thesis employs units called bunsetsu in Japanese to determine the length of direct objects in examining hypothesis 63a: that the preposed direct objects tend to be long. Hashimoto (1948) defines the bunsetsu unit as "units we first obtain by dividing sentences" and "components which immediately compose sentences" (1948: 8).3 9 For example, 64 below consists of eight bunsetsu units divided by slashes. (64) Watashi wa / kinoo / yuujin to / futari de / maruzen e / I TOP yesterday friend with two (people) by Maruzen LOC hon o / kai-ni / ikimashita. book ACC buy-purpose went 'Yesterday, I went together with my friend to Maruzen to buy a book.' (Hashimoto 1948: 6) As shown in the above example, adverbs such as kinoo 'yesterday' can themselves form bunsetsu units. On the other hand, particles and auxiliary verbs cannot by themselves form bunsetsu units; thus, they form bunsetsu units when used together with other words. Examples are watashi wa T TOP' and yuujin to 'with my friend' as shown in 64. One way to identify where a sentence is divided into bunsetsu units is to insert elements such as the sentence final particle (SFP) ne or the combination of a copula (CPL) and the SFP desu ne into the sentence (Mikami 1972) as shown in the following examples: 3 8Chapter 3 explains the methodology this thesis employs. 39The "bunsetsu unit" is equivalent to syntactic constituents such as Determiner Phrases (DPs) and verbs. I will employ the term bunsetsu unit to remain consistent with terminology employed in Saeki (1960, 1975). 46 (65) a. inserting the SFP ne: Watashi wa ne, kinoo ne, tomodachi to ne, maruzen e ne, I TOP SFP yesterday SFP friend with SFP Maruzen to SFP shomotsu o ne, . . . books ACC SFP (Mikami 1972:27) b. inserting a combination of a CPL and SFP, desu ne: Watashi wa desu ne, kinoo desu ne, tomodachi to desu ne, I TOP CPL SFP yesterday CPL SFP friend with CPL SFP maruzen e desu ne, shomotsu o desu ne, . . . Maruzen to CPL SFP books ACC CPL SFP (Mikami 1972:27) This thesis employs Hashimoto's definition and follows Mikami's method for identifying bunsetsu units. In its examination of hypothesis 63b, this thesis uses the term "demonstratives" to refer to demonstratives which are themselves nouns such as sore 'it,' as shown in 66, as well as to those modifying nouns such as sono 'that' and sonna 'that kind of,' as shown in 67. Demonstratives are presented in bold face. (66) Sore o iguchi ga uketa no da to wa. . . it ACC Iguchi NOM had NMR CPL COMP TOP 'that Iguchi had it (=an operation) . . .' (Shokujo by Takeuchi Kazuko: 315-316) (cited in Saeki 1960: 62) (67) a. Shikashi, seefu wa sono yakusoku o kutsugaeshita. however government TOP that promise ACC reversed 'However, the government reversed that promise.'40 (Jgirisu-biiki: 83) b. Sonna ree o terebi ga hookokushiteita. such example ACC television NOM was-reporting 'A television program reported such an example.' (Tenseijingo '97 aki Vol. 110: 78) The demonstrative in 66, sore 'it,' is itself a noun and is immediately followed by the particle o. The demonstratives sono 'that' in 67a and sonna 'that kind of in 67b modify the nouns yakusoku 4 0 The English translations of samples taken from the data are mine except for some from Tenseijingo '97 aki Vol. 110 and Tenseijingo '97 fuyu Vol. Ill, in which translations are provided. I modify these translations when clarification is needed. 47 'promise' and ree 'example' respectively. Saeki (1960, 1975) uses the phrase "elements containing demonstratives" in stating one of the tendencies in which "elements containing demonstratives tend to precede those not containing demonstratives" (1960: 62). Since he provides only 66 as his example, it is not clear whether he refers only to demonstratives which are themselves nouns such as sore 'it' in 66, or also to those modifying nouns such as sono 'that' and sonna 'that kind of as in 67. Based on the word "containing" in his term "elements containing demonstratives," this thesis assumes that this term refers to both demonstratives which are themselves nouns and those modifying nouns, and employs this term to refer to both types of demonstratives. The term "idiomatic expression" in hypothesis 63c refers to expressions formed with a direct object and a verb, in which the meaning of the verb differs from its original meaning as shown in the example below:41 (68) a. denpoo o utsu telegraph ACC type 'send a telegraph' b. ibiki o kaku snore ACC display 'snore' (Morita 1989: 102) The verb utsu in 74a and kaku in 68b originally mean 'type' and 'display' on their own, respectively. However, their original meanings are lost when they are combined with denpoo o 'telegraph A C C and ibiki o 'snore ACC,' expressing the idiomatic meanings of 'send a telegraph' and 'snore' respectively. This thesis employs the following diagnostic for examining whether preposing of the direct object emphasizes the direct object (hypothesis 63d). A preposed direct object is considered as emphasized when it conveys focus by corresponding to the constituent that is wh-questioned in a question (Rochemont 1998). In other words, a clause containing this.direct object is interpreted to 4'Translations in example 68 are mine. 48 convey nani o dooshita 'what someone did' as opposed to dare ga dooshita 'who did what' or nani ga dooshita 'what happened.' This thesis adopts a modified version of Prince's (1981, 1992) taxonomy of information status in order to examine the information status of the preposed direct objects (hypothesis 63e). I will not employ the notion of "activation" (cf. Chafe, 1987; Dryer, 1992; Lambrecht, 1994) because it will be difficult to assess the degree of activity of NPs in the hearer's consciousness. Shown below is the taxonomy employed in this thesis: Information status of NPs Discourse-new/ Hearer-new Unanchored Anchored Discourse-new/ Hearer-old Unused Inferrable Discourse-old/ Hearer-old Situationally Textually Evoked Evoked Figure 4. Information status of discourse entities: taxonomy of this thesis This thesis does not employ Prince's notion of the "Containing Inferrable" in Figure 4 AO since it describes the relationship between two discourse entities within an NP as shown in 69. (69) The page of that book I bought fell out. (Prince 1992: 307) In 69 the NP 'the page' is a "Containing Inferrable" because it is inferred from the NP 'that book' contained in the same NP 'the page of that book.' Since this thesis examines the relationship between preposed direct objects and their preceding discourses, and not the relationship between two entities contained in direct objects, I will not include "Containing Inferrable" in the framework of this thesis. Table 3 below shows the definition and an example of each information status. NPs 4 2The italicization is from Prince's (1992) original. 49 representing each information status are italicized in the examples. Unanchored a newly introduced ("Discourse-new/Hearer-new") NP not linked to some other discourse entity by means of another NP or 'anchor' properly contained within it. Example: I got on a bus yesterday and the driver was drunk. (Prince 1981: 233) Anchored a newly introduced ("Discourse-new/Hearer-new") NP which is linked to some other discourse entity by means of another NP or 'anchor' properly contained within it. The NP 'a guy I work with' is an "Anchored" NP containing an 'anchor,' 'I,' in the following example. Example: A guy I work with says he knows your sister. (Prince 1981: 233) Unused a newly introduced ("Discourse-new") NP which is assumed to be already known to the hearer ("Hearer-old"). Example: Noam Chomsky went to Penn. (Prince 1981: 233) Inferrable a newly introduced NP which can be inferred from "Discourse-old" entities. The NP 'the driver' is inferrable from an already mentioned NP 'a bus' in the following example. Example: I got on a bus yesterday and the driver was drunk. (Prince 1981: 233) Situationally an NP which is identified from an entity which exists in an extratextual context Evoked ("Discourse-old/Hearer-old"). For example, the NP 'you' in the following example is a "Situationally Evoked" NP which refers to a person present when this sentence was uttered. Example: Would you have a change for a quarter? (Prince 1981: 233) Textually an NP which is identified from an entity already mentioned in the preceding Evoked discourse ("Discourse-old/Hearer-old"). For example, the NP 'he' is a "Textually Evoked" NP which refers to the NP 'a guy I work with' mentioned in the preceding discourse. Example: A guy I work with says he knows your sister. (Prince 1981: 233) Table 3. Definitions of information statuses 50 This thesis employs the framework presented in Figure 4 and follows the definitions in Table 3 in its examination of the relationship between the preposed direct object and its preceding discourse. 2.3.3 Summary Section 2.3 has presented the following five hypotheses this thesis examines: (70) The direct objects in OSV clauses a. tend to be lengthy compared to those in SOV clauses; (=63a) b. tend to contain demonstratives unlike those in SOV clauses; (=63b) c. tend not to form an idiomatic expression together with a verb; (=63c) d. tend to be an emphasized NP; and (=63d) e. tend to be closely related to entities in the preceding discourse or entities in the nonlinguistic context. (=63e) This chapter has also presented the terms and taxonomy this thesis employs in its examination of the above hypotheses. 2.4 Summary This chapter reviewed previous studies of word order variations. Section 2.1 reviewed syntactic analyses that examine preposing in light of sentence grammar. Section 2.2 reviewed pragmatic analyses that examine preposing with respect to the characteristics of preposed constituents such as structural and informational. Section 2.3 presented the hypotheses and the taxonomical framework of this thesis based on the reviews in Section 2.2. The next chapter will present the methodology employed in this thesis. 51 Chapter Three Methodology This chapter explains the methodology employed to examine the hypotheses presented in Chapter 2. Section 3.1 outlines the sources of data used in this study. Section 3.2 describes what this thesis includes in and excludes from its data. Section 3.3 shows how this thesis analyzes the data. 3.1 Data Sources The data in this thesis is collected from Japanese essays written in the 1980s and 1990s by fifteen writers (eleven males and four females).1 Written materials were chosen for the following characteristics observed in written Japanese (Clancy 1988), which will be beneficial for supporting this thesis: (1) a. Particles tend not to be omitted; b. written sentences are more integrated than spoken; and c. the ellipsis of elements in written sentences occurs less frequently than in spoken. As per la, we can easily distinguish the preposed direct objects marked with the particle o from those topicalized direct objects marked with the particle wa. It would be difficult to make such a distinction in spoken Japanese since, in that language, particles are often omitted. Second, as per lb, written Japanese is the more appropriate language to study in order to examine the influence the length of direct objects has on preposing. Clancy observes that in the spoken language, relative clauses are found less often, and some noun-modifying clauses are postposed. Third, in writing, when a character in a story is mentioned for the second time, the same NPs and their pronouns that were used to introduce characters are employed, rather than having them appear as ellipses (lc). Therefore, studying the selected written characteristics will be beneficial for examining the influence of demonstratives on preposing. On the other hand, in spoken Japanese, Clancy 'Data sources are listed in Appendix A. 52 observes that the NPs used to introduce characters subsequently appear in reduced forms or as ellipses. Therefore, it would be difficult to examine the influence of demonstratives on word order variation. Furthermore, it is impossible to determine whether the direct object follows or precedes the subject when sentence elements are omitted. 3.2 Data Collection This thesis focuses on the order variation of subjects and direct objects; therefore, the collected clauses used as its data contain subjects (S), direct objects (O), and transitive verbs (V). SOV and OSV clauses which contain elements such as adverbs and indirect objects are excluded from the data. Phrases such as indirect objects and adverbs can also appear in various positions in sentences preceding verbs; therefore, it is possible for these elements to influence the order variation of subjects and direct objects. The following explains what is included and excluded from the data in terms of the subject, the direct object, and clause types. The data contain subjects marked with either the nominative case marker or the topic/contrast marker wa} Those not marked with ga or wa are not collected as data. The data also exclude SOV clauses in sentences with two subjects which are each marked by wa and clearly contrast as shown in example 2, because the preposing of direct objects does not usually occur 2The two major uses for wa are thematic (or topical) and contrastive: (i) wa for the theme of a sentence Jon wa gakusee desu. John TOP student CPL 'Speaking of John, he is a student.' (Kuno 1973: 38) (ii) wa for contrasts Ante wa futteimasu ga... rain CONT falling but -'It is raining, but . . .' (Kuno 1973: 38) Kuno (1973: 39) claims that NPs marked with the thematic wa are those which refer to "objects and concepts that have been mentioned and recorded in the registry of the present discourse" or those "in the permanent registry"—in other words, "anaphoric" NPs. On the other hand, a contrastive wa can place either "anaphoric" or "nonanaphoric" NPs in contrasting positions. 53 when two subjects are contrasted. In examples 2, 3, and 4, underlines indicate contrasted subject phrases marked with wa and capitalized words in translations indicate that these words convey contrast: (2) Mise no hoo de wa kyaku o erabenai ga, restaurant GEN side LOC CONT customer ACC cannot choose but kyaku wa jibun ni atta pabu o erabu. customer CONT oneself to suit pub ACC choose 'RESTAURANTS cannot choose their customers; however, CUSTOMERS choose the pubs which suit them.' (Igirisu-biiki: 227) What mise no hoo 'restaurants' do and kyaku 'customers' do are contrasted and both subjects precede the direct objects in example 2. The acceptability of example 2 becomes questionable when direct objects are preposed as in 3. (3) ^Kyaku o mise no hoo de wa erabenai ga, customer ACC restaurant GEN side LOC CONT cannot choose but jibun ni atta pabu o kyaku wa erabu. oneself to suit pub ACC customer CONT choose 'RESTAURANTS cannot choose their customers; however, CUSTOMERS choose the pubs which suit them.' When the direct objects kyaku 'customer' and jibun ni atta pabu 'pubs which suit them' precede the wa-marked subjects indicating contrast as in 3, direct objects are also interpreted as contrasted. Therefore, example 3 is interpreted as having two separate sets of contrasting phrases, and thus what is contrasted in this sentence is not clear. The sentence is fine when it has only one set of contrasting phrases, as in 2. Preposing either one of the two direct objects of example 2 also results in a questionable sentence as shown in 4. The first direct object, kyaku 'customer,' is preposed in 4a, and the second direct object, jibun ni atta pabu 'pubs which suit them' is 3Shibatani (1990; 265) notices that contrast is inherent in sentences such as Jon wa gakusee desu 'Speaking of John, he is a student' because "we are always isolating or singling out these propositions from other possible ones" and that contrast becomes apparent "when a parallel or contrasting proposition exists overtly or covertly." Thus, he claims that there are not two distinct was as suggested by the labels "thematic" and "contrastive," but "one and the same wa has the effect of emphasizing contrast when the discourse environment provides a background for the contrast." 54 preposed in 4b. (4) a. ^Kyaku o mise no hoo de wa erabenai ga, customer ACC restaurant GEN side LOC CONT cannot choose but kyaku wa jibun ni atta pabu o erabu. customer CONT oneself to suit pub ACC choose 'RESTAURANTS cannot choose their customers; however, CUSTOMERS choose the pubs which suit them.' b. ^Mise no hoo de wa kyaku o erabenai ga, restaurant GEN side LOC CONT customer ACC cannot choose but jibun ni atta pabu o kyaku wa erabu. oneself to suit pub ACC customer CONT choose 'RESTAURANTS cannot choose their customers; however, CUSTOMERS choose : the pubs which suit them.' Examples 4a and 4b are questionable because the preposing of a direct object causes the contrast in these sentences to become ambiguous. It seems that an OSV word order variation is not possible when a sentence contains a wa-marked subject indicating contrast followed by another wa-marked subject which is contrasted with the first subject. Therefore, sentences such as 2 are excluded from the data. The data contain direct objects marked with the particle o, which marks accusative case, and also with the dative case marker ni. The direct objects of some transitive verbs are marked with the particle ni in Japanese as shown in the below examples from Okutsu and Tanaka (1990).4 (5) a. Inu ga taroo ni kamitsuita. dog NOM Taro DAT bit 'A dog bit Taro.' b. Taroo ga hanako ni butsukatta. Taro NOM Hanako DAT ran-into 'Taro ran into Hanako.' 4The translations of examples from Okutsu and Tanaka (1990) are mine. 55 (5) c. Taroo wa hanako ni oitsuita. Taro TOP Hanako DAT caught-up-with 'Taro caught up with Hanako.' (Okutsu and Tanaka 1990: 65) This thesis does not include direct objects marked with the particle wa and those marked with the particle ga in its data. Wa-marked direct objects are excluded because they have a special function, such as topic or contrast. The underlined direct objects in examples 6 and 7 are wa-marked direct objects that comprise a topic and convey contrast, respectively. (6) Sono hon wa jon ga kaita. that book TOP John NOM wrote 'As for that book, John wrote it.' (7) Jon wa toriniku wa taberu (ga, butaniku wa tabenai ). John TOP chicken CONT eat but pork CONT doesn't-eat 'John eats chicken (but not pork).' Besides their functions, the difference between wa-marked direct objects and o-marked direct objects is the position in which they appear. O-marked direct objects can appear either in SOV or OSV word order as in 8 and 9. (8) a. Jon ga sono hon o kaita. John NOM that book ACC wrote 'John wrote that book.' b. Sono hon o jon ga kaita. that book ACC John NOM wrote 'John wrote that book.' (9) a. Jon wa toriniku o taberu John TOP chicken ACC eat 'John eats chicken (but not pork).' b. Toriniku o jon wa taberu. chicken ACC John TOP eat 'John eats chicken (but not pork).' On the other hand, wa-marked direct objects comprising a topic must precede subjects as in 6, and wa-marked direct objects indicating contrast must appear following subjects as in 7. Examples 10 56 and 11 show ungrammatical sentences in which the direct object comprising a topic follows a subject, and the contrasting direct object precedes a subject, respectively. (10) *Jon ga sono hon wa kaita. John NOM that book TOP wrote 'As for that book, John wrote it.' (11) *Toriniku wa jon wa taberu ( ga, butaniku wa tabenai ). chicken CONT John TOP eat but pork CONT doesn't-eat 'John eats chicken (but not pork).' Therefore, wa-marked direct objects are excluded from the data. Direct objects marked with the particle ga are excluded since ga-marked direct objects cannot precede subjects. Examples 12a and 12b are sentences with ga-marked direct objects from Kuno (1973), as shown by the underlines. (12) . a. Jon ga eega ga suki da. John NOM movies NOM like CPL 'John likes movies.' b. *Eega ga jon ga suki da. movies NOM John NOM like CPL 'John likes movies.' (Kuno 1973: 353) Direct objects of stative verbs, adjectives, and nominal adjectives are usually marked with the nominative case marker ga in Japanese. The nominal adjective suki 'like' is stative and the direct object eega 'movies' is marked with ga instead of o in 12. Example 12b is ungrammatical because the ga-marked direct object is preposed to the sentence-initial position and precedes the subject. NPs marked with the accusative case marker o which are not direct objects are also excluded from the data of this thesis. Examples of such NPs are those indicating a spatial path for motion verbs such as aruku 'walk' and wataru 'cross' as in 13, those indicating places departed from for motion verbs such as deru 'leave' and oriru 'descend from' as in 14, and those indicating 57 the time span of a verb such as ikiru 'live' and kurasu 'live' as in 15.5 The NPs marked with the accusative case marker o are underlined. (13) an NP indicating a spatial path: Firippo ga kogan o hashiru to, futari no musukora Philip NOM lakeshore ACC run when two (people) GEN sons to inu ga zensokuryoku de shitagau. and dog NOM full speed at follow 'When Philip runs along the lakeshore, his two sons and dog follow at full speed.' [Itaria tosukaana no yuugana shokutaku: 97) (14) an NP indicating a place departed from: Yagate, watashi ga okkusufoodo daigaku o saru hi ga before long I NOM Oxford university ACC leave day NOM chikaduitekita. came near 'Before long, the day I would leave Oxford University drew near.' (Igirisu-biiki: 16) (15) an NP indicating a time span: Watashi wa garantoshita ryoo no is-shitsu de, I TOP empty-looking dormitory GEN one-NC LOC samui wabishii mikkakan o sugoshita. cold miserable three days ACC spent T spent three cold and miserable days in the empty-looking dormitory room.' (Saigo no hanadokei: 14) SOV and OSV clauses were collected as data from main clauses, subordinate clauses, and noun-modifying clauses. A main clause is a clause which can form a sentence independently. A subordinate clause cannot stand by itself, but can help to make a larger clause when it is part of, or joined to, a main clause. Example 16 consists of one clause, the main clause. 5Martin (1975) calls an o-marked NP indicating places traversed a "traversal" object, an NP indicating places departed from an "ablative" object, and an NP indicating time spent a "temporal" object. 58 (16) Daigakuinsee da to yuu hitori no wakai josee ga graduate student CPL QT say one (person) GEN young female NOM te o agete-kureta. hand ACC raise-gave (Lit.)'One young woman who is said to be a graduate student did me the favour of raising her hand.' 'One young woman who is said to be a graduate student raised her hand.' (Uwanosora: 67) Example 17 below demonstrates a subordinate clause: (17) Watashi ga jibun mo katsute nyuu yooku ni sundeita koto o I NOM oneself also once New York in lived NMR ACC hanasu to , okusan wa me o kagayakaseta. tell when wife TOP eyes ACC brightened 'When I told (her) that I had also lived in New York once, the wife brightened her eyes.' (Itaria tosukaana no yuugana shokutaku: 182) There are two clauses in 17: a subordinate clause and a main clause. Each clause has its predicate and subject, hanasu 'talk' and watashi T in the subordinate clause and kagayakaseta 'brightened' and okusan 'wife' in the main clause. This thesis excludes, however, sentences in which the predicate of a subordinate clause and a main clause share the same subject as in example 18: (18) Tooma ga, minna no gurasu ni wain o sosogi-nagara Toma NOM everyone GEN glass into wine ACC pour-while watashi o mita. I ACC looked 'Toma looked at me while he was pouring wine into everyone's glass.' (Itaria tosukaana no yuugana shokutaku: 71) Tooma ga 'Toma NOM' in 18 is the subject phrase of the predicate of the underlined subordinate clause minna no gurasu ni wain o sosogu 'pouring wine into everyone's glass,' and at the same time is also the subject of the predicate watashi o mita 'looked at me' in the main clause. Therefore, cases such as 18 are excluded from the data. Example 19 below demonstrates a noun-modifying clause: 59 (19) Ningen ga ningen o hyookasuru toyuu koto wa hontooni human NOM human ACC evaluate COMP fact TOP truly muzukashiku, osoroshii koto da. difficult terrible thing CPL 'The fact that one human evaluates another is a truly difficult and terrible thing.' (Yotsuba no essei: 232) The noun koto 'fact' is modified by a clause and together they form an NP in 19. A total of 311 SOV and OSV clauses, falling under the categories described in this section, are collected from 1454 pages of essays written in Japanese. The next section describes how this thesis analyzes the collected 311 clauses. 3.3 Data Analyses Saeki (1960) and Miyajima (1964) analyze word order in two separate sets of data in their studies. One of their sets contains NPs modifying verbs that a) consist of the same number of bunsetsu units; b) do not contain demonstratives; c) are not followed by kakari-joshi; d) are independent from each other in their meanings; and e) are independent in their meanings when they immediately precede the verb (i.e. Idiomatic expressions).6 The other set contains those that do not fall under the categories of the first set. The main objective of Saeki's study is to determine the basic word order of Japanese, and that of Miyajima's is to examine the frequency of various word orders in an environment when influence from structural characteristics is minimized. Therefore, Saeki and Miyajima analyze their data under two separate sets. The objective of this study, in contrast, is to examine how certain characteristics of direct objects affect word order. Therefore, unlike Saeki and Miyajima, the data in this thesis are not separated into two different sets. This thesis employs quantitative analyses when examining the distribution of SOV and OSV word orders a) in data as a whole; b) in relation to the length of direct objects; c) in clauses with direct objects containing demonstratives; d) in relation to the direct object forming an idiomatic expression together with a verb; and e) in relation to the information status of direct objects. It also 6See Appendix B for definitions of bunsetsu unit, demonstrative, kakari-joshi, and idiomatic expression. 60 employs qualitative analyses for the further examination of the data. 61 Chapter Four Analyses of Data This chapter analyzes 311 clauses collected as data following the method described in Chapter 3. Section 4.1 examines the distribution of SOV and OSV word orders throughout the data. This section further discusses SOV and OSV word orders in the various environments from which the data was taken. Sections 4.2 to 4.6 each examines the following five hypotheses already presented in Chapter 2: (1) The direct objects in OSV clauses a. tend to be lengthy compared to those in SOV clauses; b. tend to contain demonstratives unlike those in SOV clauses; c. tend not to form an idiomatic expression together with a verb; d. tend to be an emphasized NP; and e. tend to be closely related to entities in the preceding discourse or entities in the nonlinguistic context. Section 4.7 summarizes the results of the analyses in Sections 4.2 to 4.6. 4.1 Distribution of SOV and OSV Clauses The following illustrates clauses with SOV and OSV word order. Sentences 2 and 3 are examples of SOV and OSV clauses respectively.1 (2) Kanojo wa nanakakokugo o hanasu. she TOP seven-languages ACC speak 'She speaks seven languages.' (Uwanosora: 145) (3) Sonna ree o terebi ga hookokushiteita. such example ACC television NOM was-reporting 'A television program reported such an example.' (Tenseijingo '97 aki Vol. 110: 78) Table 4 shows the distribution of all SOV and OSV clauses in the total of 311 clauses collected and lrThe direct object phrases of SOV and OSV clauses are underlined throughout the data analyses. 62 analyzed in this thesis.2 SOV OSV Total Number of clauses collected 267 85.9% 44 14.1% 311 100% x2 =159.9, df=l,p<0.05 Table 4. Distribution of SOV and OSV clauses Out of the 311 clauses, 85.9% (267/311) have SOV word order and 14.1% (44/311) have OSV word order.3 The frequency of the occurrence of SOV to OSV in this study is approximately 6.1 to 1 (267/44). The data contain SOV and OSV clauses collected from seven books. Table 5 below shows the distribution of SOV and OSV word orders in each book listed in alphabetical order.4 2Unless noted, this thesis employs chi-square test or chi-square test of association, which is used to determine whether two or more categorical variables are related or whether they are independent, in order to statistically verify the relationship between row and column variables of tables throughout the data analyses. 3Miyajima (1964) collected a total of 1365 clauses, each containing a subject, direct object, and verb, and finds 1307 (95.8%) clauses with the SOV word order and 58 (4.2%) clauses with the OSV word order. This thesis does not compare its data with that of his study for the following three reasons: First, the methodology employed in his study differs from that of this thesis. Miyajima separated his data into two groups: the first group excludes NPs which a) contain a different number of bunsetsu units b) contain demonstratives; and c) are accompanied by kakari-joshi such as shika 'only' and mo 'also,' and the second group contains all NPs including those excluded from the first group. Second, the amount of data collected in Miyajima's study is much greater than that of this study; therefore, it is difficult to compare the two results quantitatively. Third, Miyajima does not provide qualitative analyses in his study. 4The underlined percentages indicate that of the row categories (the distribution of SOV and OSV clauses in each environment), and the percentages in parentheses indicate that of the column categories (the distribution of each environment in SOV and OSV clauses) in Tables 5 to 14 except for Table 8. 63 SOV OSV Total Igirisu-biiki 23 95.8% 1 4.2% 24 100% (8.6%) (2.3%) (7.7%) Itaria tosukaana no 28 90.3% 3 9.7% 31 100% yuugana shokutaku (10.5%) (6.8%) (10.0%) Saigo no hanadokei 48 72.7% 18 27.3% 66 100% (18.0%) (40.9%) (21.2%) Tenseijingo '97 aki 49 90.7% 5 9.3% 54 100% Vol. 110 (18.3%) (11.4%) (17.4%) Tenseijingo '97 fuyu 37 100% 0 0% 37 100% Vol. Ill (13.9%) (0%) (11.9%) Uwanosora 68 84.0% 13 16.0% 81 100% (25.5%) (29.5%) (26.0%) Yotsuba no essei 14 77.8% 4 22.2% 18 100% (5.2%) (9.0%) (5.8%) Total 267 85.9% 44 14.1% 311 100% (100%) (100%) (100%) Table 5. Distribution of SOV and OSV clauses in the seven source books The frequency of the use of OSV word order varies between books as shown in Table 5. All the books contain both SOV and OSV clauses except for Tenseijingo '97 fuyu Vol. Ill, in which no OSV clause was found. It should be noted that though the same group of authors wrote both Tenseijingo '97 fuyu Vol. Ill and Tenseijingo '97 aki Vol. 110, the latter work contains five OSV clauses. It should also be noted that Igirisu-biiki is a collection of ten short essays each written by a different author. Some books contain a higher percentage of OSV clauses than others 64 do. For example, Saigo no hanadokei contains approximately 41% of the forty-four OSV clauses in the data, and Uwanosora contains approximately 30%. This may be due to the individual styles of the authors. The data contain SOV and OSV clauses taken from three types of clauses, with their subjects marked either with the nominative case marker ga or the topic/contrast marker wa. The following subsections examine the distribution of SOV and OSV clauses in these environments. Section 4.1.1 discusses SOV and OSV word orders in the three types of clauses. Section 4.1.2 discusses SOV and OSV word orders in clauses with subjects marked with the nominative case marker ga and the topic/contrast marker wa. 4.1.1 SOV and OSV Word Orders in Three Types of Clauses For this thesis, SOV and OSV clauses were collected from main clauses, subordinate clauses, and noun-modifying clauses. Examples 4, 5, and 6 show SOV clauses, and 7, 8, and 9 show OSV clauses taken from a main clause, a subordinate clause, and a noun-modifying clause respectively. (4) a main clause with SOV word order Daigakuinsee da to yuu hitori no wakai josee ga graduate student CPL QT say one (person) GEN young female NOM te o agete-kureta. hand ACC raise-gave (Lit.)'One young woman who is said to be a graduate student did me the favour of raising her hand.' 'One young woman who is said to be a graduate student raised her hand.' (Uwanosora: 67) 65 (5) a subordinate clause with SOV word order Watashi ga jibun mo katsute nyuu yooku ni sundeita koto o I NOM oneself also once New York in lived NMR ACC hanasu to, okusan wa me o kagayakaseta. tell when wife TOP eyes ACC brightened 'When I told (her) that I had also once lived in New York, the wife brightened her eyes.' (Itaria tosukaana no yuugana shokutaku: 182) (6) a noun-modifying clause with SOV word order Ningen ga ningen o hyookasuru toyuu koto wa hontooni human NOM human ACC evaluate COMP fact TOP truly muzukashiku, osoroshii koto da. difficult awful thing CPL 'That one human evaluates another is a truly difficult and awful thing.' (Yotsuba no essei: 232) (7) a main clause with OSV word order Nani o kare wa tsutaetakatta no ka. what ACC he TOP wanted-to-convey (a message) NMR Q '(I wonder) what message he wanted to convey?' (Tenseijingo '97 aki Vol. 110: 75) (8) a subordinate clause with OSV word order Katsute furankii sakai ga netsuenshite. koohyoo once Franky Sakai NOM give-an-ardent-performance favourable-comment o hakushita 'watashi wa kai ni naritai' o ACC received I TOP shellfish into want-to-become ACC tokoro jooji ga enjite, futatabi 'watashi wa kai ni Tokoro George NOM perform again I TOP shellfish into naritai' ga terebi de hoosoosareta. want-to-become NOM television on was-broadcasted 'George Tokoro performed "I want to become a shellfish," of which Franky Sakai once gave an ardent performance and received favourable comments, and "I want to become a shellfish" was broadcast on television again.' (Saigo no hanadokei: 146) 66 (9) a noun-modifying clause with OSV word order Ichioku-en ijoo no okane o, futsuu no hito one hundred million-NC greater GEN money ACC ordinary GEN people ga soozoodekinai no to sore wa niteiru. NOM cannot-imagine NMR to it TOP similar 'It is similar to [the fact] that ordinary people cannot imagine an amount of money that is greater than one hundred million yen.' (Yotsuba no essei: 12) The distribution of the SOV and OSV word orders in all three types of clauses is analyzed in order to determine whether the OSV word order appears in different ratios according to the type of clause. Table 6 shows the distribution of the SOV and OSV word orders in main, subordinate, and noun-modifying clauses. SOV OSV Total Clause types Main clause 160 85.6% 27 14.4% 187 100% (59.9%) (61.4%) (60.1%) Subordinate clause 49 84.5% 9 15.5% 58 100% (18.4%) (20.4%) (18.6%) Noun-modifying clause 58 87.9% 8 12.1% 66 100% (21.7%) (18.2%) (21.2%) Total 267 85.9% 44 14.1% 311 100% (100%) (100%) (100%) x2=0.33, df=2, NS (p>0.05) Table 6. Distribution of main, subordinate, and noun-modifying clauses SOV and OSV word orders in main, subordinate, and noun-modifying clauses are observed to have similar frequencies to those in the data as a whole, as shown in Table 6. The distributions of the three types of clauses, indicated by the percentages in parentheses, do not differ significantly 67 between the SOV and OSV word orders. Furthermore, the chi-square test of association indicates that word order and clause types are not related. The results shown in Table 6 suggest that, as far as this study is concerned, the types of clauses sampled do not affect the ratio of the SOV word order to the OSV. 4.1.2 SOV and OSV Word Orders and Subjects Marked with Ga and Wa The data from this study contain both subjects marked with the nominative case marker ga and the topic/contrast marker wa. Sentences 10 and 11 show examples of clauses with subjects marked with ga in SOV and OSV word orders, and 12 and 13 are clauses with subjects marked with wa in SOV and OSV word orders respectively. (10) an SOV clause with a ga-marked subject Daigakuinsee da to yuu hitori no wakai josee ga graduate student CPL QT say one (person) GEN young female NOM te o agete-kureta. hand ACC raise-gave (Lit.)'One young woman who is said to be a graduate student did me the favour of raising her hand.' 'One young woman who is said to be a graduate student raised her hand.' (Uwanosora: 67) (11) an SOV clause with a wa-marked subject Karera futari wa onajiyoona tsuiido no fuyu-gaitoo o they two (persons) TOP similar tweed GEN Winter-coat ACC kite it a. were-wearing 'The two of them were wearing similar tweed winter coats.' (Igirisu-biiki: 33) 68 (12) an OSV clause with a ga-marked subject Dosekiryuu de giseesha ga deta kagoshima-ken no landslide because-of victim NOM be caused Kagoshima-prefecture GEN genchi o kamei shizuka kensetsu-soo ga shisatsushita. site ACC Kamei Shizuka Construction-Minister ACC visited 'Construction Minister Shizuka Kamei visited the site in Kagoshima Prefecture where many people had been killed in a landslide.' (Tenseijingo '97 aki Vol. 110: 24) (13) an OSV clause with a wa-marked subject Nimokakawarazu "saibansho ga ninteeshita hanz.ai-jijitsu o nevertheless court NOM established criminal-facts ACC watashi wa mitomenai" nado to kyoobenshita. I TOP do-not-acknowledge EMPH QT asserted 'Nevertheless, he asserted, "I do not acknowledge the criminal facts established by the court.'" (Tenseijingo '97 aki Vol. 110: 133) The particle ga marks the subjects of sentences in either a neutral description or an exhaustive listing, and the particle wa marks those which are either thematic or contrastive (Kuno 1973).5 Therefore, sentences 10 and 11, whose subjects are marked with ga, are interpreted as neutral descriptions or exhaustive listings. Wa-marked subjects in 12 and 13 are interpreted as either topic or contrast subjects.6 Since the functions of ga-marked and wa-marked subjects are different, I will examine the relationship between the distributions of the SOV and OSV word orders and subjects marked with ga and wa. Table 7 shows the distribution of SOV and OSV clauses with ga-marked and wa-marked subjects. 5See Kuno (1973) for a further discussion on subjects marked with ga and wa. 6 Although the role of in 10 and 12 can be translated, in a certain context, as an exhaustive listing, I chose to translate them as the agents of sentences of neutral description because the contexts in which these sentences appear give neutral description interpretations. Subjects in 11 and 13 are translated as topics from the contexts of these sentences. 69 SOV OSV Total Subjects ga-marked 150 86.7% 23 13.3% 173 100% (56.2%) (52.3%) (55.6%) wa-marked 117 84.8% 21 15.2% 138 100% (43.8%) (47.7%) (44.4%) Total 267 85.9% 44 14.1% 311 100% (100%) (100%) (100%) x2=0.23, df=l, NS (p>0.05) Table 7. Distribution of ga-marked and wa-marked subjects The differences in distribution of the SOV and OSV word orders in clauses with subjects marked with ga and wa do not differ significantly. This suggests that direct objects can precede both ga-marked and wa-marked subjects, and that the OSV word order can be observed as having similar frequencies in clauses with subjects marked with two different particles. Table 7 also shows that the SOV and OSV clauses in this study's data contain approximately the same percentage of ga-marked and wa-marked subjects. In both SOV and OSV clauses, a little more than half of the subjects are marked with ga. The results shown in Table 7 suggest that the two particles marking the subjects are not related to the distribution of the SOV and OSV clauses in this study. Section 4.1 examined the distributions of SOV and OSV word orders in the data as a whole, in three different types of clauses, and in clauses with subjects marked by two different particles. We observed that neither the three types of clauses the data were taken from, nor the two particles marking the subjects ga and wa affect the distribution of SOV or OSV word order. Therefore, I will not specify types of clauses and particles marking subjects when examining the five hypotheses presented in 1 in the following sections. 70 4.2 Length of Direct Objects This section focuses on the first hypothesis concerning the length of direct objects. As explained in Chapter 2, subjects and direct objects are divided into bunsetsu units, which are counted in order to determine the length of the subjects and direct objects, and are examined to ascertain the influence their length has on word order variation.7 Direct objects in OSV clauses contain more bunsetsu units, on average, than direct objects in SOV clauses. Table 8 shows the average number of bunsetsu units in the subjects (S) and direct objects (O) of the collected SOV and OSV clauses.8 SOV OSV Total Number of clauses collected 267 44 311 Total number of bunsetsu units in S 515 69 584 Total number of bunsetsu units in 0 726 228 954 Average number of bunsetsu units in Sa) 1.93 1.57 1.88 Average number of bunsetsu units in Ob) 2.72 5.18 3.07 a) t = 1.823, df = 300, p<0.05 b) t= 5.334, df = 300, p<0.05 Table 8. Average number of bunsetsu units in S and O The direct objects in OSV clauses contain approximately twice the number of bunsetsu units as direct objects in SOV clauses; the average number of bunsetsu units contained in direct objects in OSV and SOV clauses is 5.18 and 2.72 respectively. On the other hand, the lengths of subjects in SOV and OSV clauses do not differ significantly; the average number of bunsetsu units contained 7See appendix B for the definition of a bunsetsu unit. 8 This thesis employs the f-test, used to determine whether two group means differ significantly, in order to statistically verify the results in Table 8. 71 in the subjects is 1.93 in SOV clauses and 1.57 in OSV clauses. The results in Table 8 suggest that the direct objects in OSV clauses are longer than those in SOV clauses. Table 9 shows the distribution of SOV and OSV clauses in relation to the number of bunsetsu units contained in subjects (S) and direct objects (O). The distribution of SOV and OSV clauses is examined in the following three environments in which a) subjects contain a greater number of bunsetsu units than direct objects (S>0); b) subjects and direct objects contain the same number of bunsetsu units (S=0); and c) direct objects contain a greater number of bunsetsu units (S<0). SOV OSV Total Number of bunsetsu units S>0 65 94.2% 4 5.8% 69 100% (24.3%) (9.1%) (22.2%) S=0 85 90.4% 9 9.6% 94 100% (31.8%) (20.4%) (30.2%) S<0 117 79.1% 31 20.9% 148 100% (43.8%) (70.4%) (47.6%) Total 267 85.9% 44 14.1% 311 100% (100%) (100%) (100%) x 2 = 11.30, df=2, p<0.05 Table 9. Distribution of SOV and OSV clauses in relation to the number of bunsetsu units contained in S and O The OSV word order is observed more in the environment S<0. In the environments S>0 and S=0, 5.8% and 9.6% of all the clauses, respectively, follow the OSV word order. The frequency of the OSV word order increases to 20.9% in the environment S<0. Therefore, Table 9 suggests that the OSV word order occurs more frequently when direct objects contain more bunsetsu units. 72 Among the 267 SOV clauses, a little less than half (43.8%) of all direct objects contain a greater number of bunsetsu units than subjects. In contrast, 70.4% of direct objects in the forty-four OSV clauses contain more bunsetsu units than subjects. This indicates that the direct objects in OSV clauses tend to contain more bunsetsu units, and are thus longer, than those in SOV clauses. On the other hand, some SOV clauses in the data contain lengthy direct objects as shown in examples 14 to 17.9 (14) Neru ga, / mura no / noofu ga / ishi no / oobun de / Nel NOM village GEN farmer NOM stone GEN oven in yaita toyuu, / kiji ga / araku. / soredeite / shhtorishita / baked COMP dough NOM coarse but moist ajiwai no / pan o / kiri-waketa. flavour GEN bread ACC cut-divided 'Nel cut and divided the bread, which village farmers baked in a stone oven, and whose dough was coarse but had a moist flavour.' (Itaria tosukaana no yuugana shokutaku: 71) (15) Tooma to /Neru wa, / jenova no / chuushin-gai de / keeeesuru / Toma and Nel TOP Genova GEN centre-street LOC manage garoo no / hokani / ni-ken no / apaato to / koogai no / art gallery GEN other than two-NC GEN apartment and suburbs GEN bessoo to / gookee / yon-ken mo no / ie o / motsu. villa COMP total four-NC EMPH GEN house ACC have 'Toma and Nel have, other than the art gallery that they manage in the main street in Genova, two apartments and a villa in the suburbs: as many as four houses in total.' (Itaria tosukaana no yuugana shokutaku: 114) 9The slashes in examples 14 to 17 indicate where these clauses are divided into bunsetsu units. 73 (16) Wareware wa, / kyoo. / kono / shiboo-ran ni / notta / we TOP today this death-column LOC appeared hitotachi ga. / hanavakana / seeshun ya / soonen o / people NOM bright youth and in one's mature age ACC okutta ato. / kodokude / minikui / ichi, ni nen o / spent after lonely ugly one two year ACC sugoshita koto o / shitteiru. spent NMR ACC know 'We know that the persons whose names appeared today in this obituary column spent a couple of lonely ugly years after experiencing the brightness of youth and their mature age.' (Saigo no hanadokei: 191) (17) Watashi wa / E. M. fosutaa no / shoosetsu ni / tooioosuru / I TOP E. M. Forster GEN novel in appear eekokuiin ga. / itaria o / otoz.ureteiru / aida ni / sono / British person NOM Italy ACC visiting while in that seerai no / kingen-iicchokuna / moraru ga / ayafuya ni / natural GEN serious-faithful moral NOM uncertain DAT nari, / itsushika / itaria no / amooru / afureru / become before-one-is-aware Italy GEN amour be-full kibun ni / somatte / 'myoona' / kimochi ni / natteiku / feeling DAT be-influenced strange feeling DAT become monogatari o / omoidashiteita. story ACC was-recalling T recall a story in which an Englishman, who appeared in a novel by E.M. Forster, developed a strange feeling while visiting Italy, after being influenced unawares by an overpowering Italian sense of "amour" and losing all natural sense of morals.' (Itaria Tosukaana no yuugana shokutaku: 107) Three of the examples given above, 14, 15, and 17, are taken from Itaria tosukaana no yuugana shokutaku, where it should be noted there are also some OSV clauses with lengthy direct objects. The data contain a total of thirty-one clauses from this book, twenty-eight in SOV word order and three in OSV word order.10 Examples 18, 19, and 20 are OSV clauses also taken from Itaria tosukaana no yuugana shokutaku. 10See Table 5 for the distributions of SOV and OSV word orders in this study's seven source books. 74 (18) Shio-aii no / nai / tosukaana-pan no / buaisoo-buri o / salt-taste GEN not exist Tuscany-bread GEN unsociable-manner ACC shita ga oboeteite, hito-kuchi kamu tabini, ichi-nen-mae no tongue NOM remembering one-bite bite every time one-year-ago GEN taiyoo to shoka no midori ga karada-juu ni yomigaeru. sun and early summer GEN green NOM body-all DAT come back 'My tongue remembers the unsociable manner of Tuscany bread, which does not have a salty flavour, and every time I take a bite, the sun and the green of the early summer of a year ago come back to all parts of my body(7ton'a tosukaana no yuugana shokutaku: 138) (19) Dooyara, honmono no / yasai o / honkide / tarafuku / taberu / it-seems real GEN vegetables ACC seriously heartily eat kyaku ga / iru / koto o /. nihon no shefu wa customer NOM exist NMR ACC Japan GEN chef TOP gozonjinai no dewanai ka. do-not-know NMR CPL-neg Q 'It seems that chefs in Japan do not know that there are customers who eat real vegetables seriously and heartily.' (Itaria tosukaana no yuugana shokutaku: 36) (20) Soo ii-nagara, watashi wa sooitta / taishuu / shoohi / bunka no / so say-while I TOP such popular consumption culture GEN kataboo o /. kyuusokuni yutaka ni natta nihonjin to (take) part (in) ACC rapidly rich to became Japanese people and japan-manee ga katsuideiru koto ni omoi-itatta ga, Japanese money NOM taking (part) in NMR to think-reached but sore wa iwazuni oita. it TOP without-saying left 'While saying so, I thought of the fact that Japanese people who have become rich rapidly and Japanese money have contributed to such popular consumption culture; however, I did not mention it.' (Itaria tosukaana no yuugana shokutaku: 75) The above examples 14 to 20 show that the length of direct objects does not fully explain the OSV word order: the direct objects in 18, 19, and 20 precede the subjects although they contain a smaller number of bunsetsu units than the direct objects in 14, 15, and 17, which follow the SOV word order. Notice that the direct object phrase in example 20 kataboo o forms an idiomatic expression 'taking part in' together with the verb katsuideiru. Section 4.4 discusses the relationship between the OSV word order and idiomatic expressions. 75 The differences between the above SOV and OSV clauses can be explained by examining them in context. In the text preceding the above OSV clauses 18, 19, and 20 cited above, the storyline centres around their direct objects. For example, the author writes about the dinner she had that night, which includes tosukaana-pan 'Tuscany bread' prior to 18. In the text preceding 19, she writes about how she enjoys having fresh vegetables in Italy and how disappointed she is with the salad they serve in restaurants in Japan, and in the discourse prior to 20 how what used to be enjoyed only by the privileged class has been popularized. On the other hand, the direct objects in SOV clauses 14, 15, and 17 are not central to the author's narrative, though they are related. For example, before 14 the author writes about a big platter filled with vegetables and cheese. She touches on the bread in 14, and then returns to the contents of the platter. Though written by a different author, the same can be said for example 16. In the discourse preceding 16, he writes about what people of his sexagenarian age group are interested in and their thoughts on various things, but only touches on people who appear in the obituary column. Sections 4.5 and 4.6 further discuss the relationship between the information the direct object conveys and OSV word order. As we saw in Chapter 2, Saeki (1960, 1975) claims that long elements, particularly those containing clauses, tend to precede others.11 Table 10 shows the distribution of SOV and OSV 1 'The following examples illustrate long direct objects containing (i) a clause or (ii) no clause: (i) Un'yushoo no choosa iinkai ga, fukuoka kuukoo de ministry of transportation G E N investigation committee N O M Fukuoka airport L O C okotta indoneshia-ki enioo-jiko no gen'in o happyooshita. happened Indonesia-airplane fire-accident G E N cause A C C announced 'The Ministry of Transportation investigation committee announced the cause of the Indonesian plane crash that happened at Fukuoka Airport.' (Tenseijingo '97 fuyu Vol. Ill: 95) (ii) Kashi-gumi yuushoo no kan'emon wa, maniuu goiuk-ko. yookan sweets-division champion N M R Kan'emon TOP manjuu fifty-NC yookan nana-sao. usukawamochi sanjuk-ko o tairageta. seven-NC usukawamochi thirty-NC A C C ate up 76 clauses which have direct objects containing clauses ("+ Clause"), and direct objects without clauses ("- Clause"). SOV OSV Total Direct objects + Clause 64 71.1% (24.0%) 26 28.9% (59.1%) 90 100% (28.9%) • - Clause 203 91.9% (76.0%) 18 8.1% (40.9%) 221 100% (71.1%) Total 267 85.9% (100%) 44 14.1% (100%) 311 100% (100%) x2=22.67, df=l, p<0.05 Table 10. Distribution of "+ Clause" and "- Clause" direct objects Out of the ninety clauses containing "+ Clause" direct objects, twenty-six (28.9%) had the OSV word order and sixty-four (71.1%) had the SOV word order. On the other hand, the "- Clause" environment observed the OSV word order less frequently; out of the 221 clauses containing "-Clause" direct objects, only eighteen (8.1%) had the OSV word order and 203 (91.9%) had the SOV word order. It should be noted that NPs tend not to contain clauses when they contain demonstratives due to the pronominal characteristics of the demonstratives. Therefore, it is not surprising that 40.9% (18/44) of the direct objects in OSV clauses are "- Clause." I will discuss the relationship between the direct objects containing demonstratives and OSV word order in Section 4.3. The results in Table 10 show that direct objects are more likely to precede subjects when direct objects contain clauses, thus supporting Saeki's claim that long elements, particularly clauses, tend to precede others. Section 4.5 further discusses the relationship between clauses and 'Kan'emon, who was the champion of the sweets division, polished off fifty manjuu [buns filled with bean-jam], seven bars of yookan [sweet bean jelly], and thirty usukawamochi [rice cakes with thin coverings].' (Tenseijingo '97 aki Vol. 110: 14-15) 77 clauses, tend to precede others. Section 4.5 further discusses the relationship between clauses and word order. This section examined the influence the length of direct objects has on the OSV word order. We observed the following four characteristics of subjects and direct objects in relation to their length: (i) Direct objects in OSV clauses contain on the average a larger number of bunsetsu units than direct objects in SOV clauses (Table 8). (ii) Subjects tend to contain a larger number of bunsetsu units in SOV clauses than in OSV clauses (Table 8). (iii) OSV clauses tend to have more direct objects, containing a larger number of bunsetsu units than subjects, than do SOV clauses (Table 9). (iv) OSV clauses occur more frequently in the environment where direct objects contain clauses (Table 10). The above findings support the tendency suggested in Saeki (1960, 1975) for long elements to precede others. However, we have also seen that there are cases in which long direct objects do not precede subjects, as shown in examples 14 to 17. Furthermore, although OSV clauses occur more frequently when direct objects are long, the frequency of the occurrence of OSV clauses is still low compared to that of SOV clauses. Therefore, the length of direct objects influences word order and to some extent creates OSV clauses; however, this is not the only factor which decides word order. 4.3 Demonstratives in Direct Objects This section examines the second hypothesis of this thesis, which concerns the influence the appearance of demonstratives in direct objects has on word order variation. Demonstratives which are themselves nouns such as sore 'it,' and demonstratives modifying nouns such as sono 'that' and sonna 'that kind of are included in the data as explained in Chapter 3. The following sentences 21 and 22 show examples of a demonstrative that is itself a noun and a demonstrative modifying a noun, respectively. Demonstratives are presented in bold face. 78 (21) Senpai kangofu ga sore o hagemashiteiru. senior nurse NOM it ACC encouraging 'A senior nurse encourages it (=a new nurse crying).' (Saigo no hanadokei: 131) (22) Sonna ree o terebi ga hookokushiteita. such example ACC television NOM was-reporting 'A television program reported such an example.' {Tenseijingo '97 aki Vol. 110: 78) Table 11 illustrates the number of direct objects containing demonstratives in SOV and OSV clauses. In the table, direct objects which contain demonstratives are shown as "+ Demonstrative" and direct objects which do not contain demonstratives are shown as "- Demonstrative." SOV OSV Total Direct Objects -(-Demonstrative 29 70.7% 12 29.3% 41 100% (10.9%) (27.3%) (13.2%) -Demonstrative 238 88.1% 32 11.9% 270 100% (89.1%) (72.7%) (86.8%) Total 267 85.9% 44 14.1% 311 100% (100%) (100%) (100%) x2=8.89, df=l, p<0.05 Table 11. Distribution of "+ Demonstrative" and "- Demonstrative" direct objects OSV clauses are more likely to contain "+ Demonstrative" direct objects than SOV clauses. There are 10.9% "+ Demonstrative" direct objects in SOV clauses and 27.3% in OSV clauses. Although only twelve (29.3%) of the forty-one "+ Demonstrative" direct objects in the data occur in OSV clauses, the frequency of occurrence of OSV clauses in the "+ Demonstrative" environment is approximately two times more than that in the data as a whole, which is 14.1%, as shown in Table 11. Therefore, the results in Table 11 suggest that OSV clauses occur more frequently when direct 79 objects contain demonstratives. As shown in Table 11, there are forty-one direct objects in the data containing demonstratives. Among the forty-one "+ Demonstrative" direct objects, there are some "+ Demonstrative" direct objects which themselves do not refer to a person or object. The following sentence 23 shows an example of such a "+ Demonstrative" direct object. (23) Zakusenhauzen o sagasu no ni wa kurooshita. Sachsenhausen ACC find NMR in TOP went through difficulties Nishi-berurin ni nijyuu-nen sundeiru doitsujin no yuujin wa West Berlin LOC twenty-year live German GEN friend TOP sono sonz.ai o shirazu, infomeeshon sentaa ni itte kiite-mo its existence ACC ., don't-know information centre to go-and ask-even yooryoo-o-enai. be-off-the-point T went through some difficulties in finding Sachsenhausen. A German friend, who has been living in West Berlin for twenty years, did not know of its existence, and even when I went and asked at the information centre, their information was unclear.' (Uwanosora: 126) The demonstrative sono 'its,' in the direct object sono sonzai 'its existence,' refers to Zakusenhauzen 'Sachsenhausen' in the preceding sentence. However, the direct object sono sonzai 'its existence' itself does not refer to an entity in the preceding discourse. In this respect, the "+ Demonstrative" direct object in 23 is different from those in 21 and 22. The following is another example of a "+ Demonstrative" direct object which itself does not refer to an entity in the preceding discourse. (24) Sonna shakaiteki jookyoo o mushishite ichi-oku o such social circumstance ACC ignore one-hundred million ACC koeru kane o hirooen ni kakeru shinkee o watashi wa exceed money ACC reception on spend nerve ACC I TOP utagau kara dearu. mistrust because CPL 'It is because I mistrust the nerve that ignores such social circumstances and spends an amount of money exceeding one hundred million yen on a wedding reception.' (Saigo no hanadokei: 31) 80 The direct object in 24 contains a demonstrative sonna 'such' in sonna shakaiteki jookyoo 'such social circumstances' which refers to the circumstance described in the preceding discourse. The phrase sonna shakaiteki jookyoo 'such social circumstances' is a part of the analyzed direct object sonna shakaiteki jookyoo o mushishite ichi-oku o koeru kane o hirooen ni kakeru shinkee 'the nerve that ignores such social circumstances and spends an amount of money exceeding one hundred million yen on a wedding reception,' and this direct object itself does not refer to a person or an object in the preceding discourse, unlike the direct objects in examples 21 and 22. Example 24 is also different from 23 in that the NP modified by the demonstrative sonna 'such,' shakaiteki jookyoo 'social circumstances' is itself not the direct object of the main clause being analyzed. Table 12 shows the distribution of "+ Demonstrative" direct objects which themselves refer to an entity in discourse ("Type A"), such as those in 21 and 22, and those which themselves do not refer to an entity in discourse ("Type B"), such as those in 23 and 24, in the data. SOV OSV Total + Demonstrative Type A 27 77.1% 8 22.9% 35 100% (93.1%) (66.7%) (85.4%) Type B 2 33.3% 4 66.7% 6 100% (6.9%) (33.3%) (14.6%) Total 29 70.7% 12 29.3% 41 100% (100%) (100%) (100%) x2=4.73, df=l, p<0.05 Table 12. Distribution of Type A and Type B "+ Demonstrative" direct objects The distribution of SOV and OSV word orders is reversed among "Type B + Demonstrative" direct objects. This may be because three out of the four "Type B" direct objects in OSV clauses are 81 quite lengthy, as in the previously shown example 24. Example 25 below also contains a lengthy "Type B" direct object. (25) Osoraku. ooku no hito ga shiranu kono shoosasshi ga probably many GEN people NOM do-not-know this booklet NOM ima no wakai sedai ni yomareru koto o watashi wa now GEN young generation by read NMR ACC I TOP negatte-yamanai. hope-cannot-stop T cannot stop hoping that this booklet, which many people probably do not know about, will be read by the young generations of today.' (Saigo no hanadokei: 96) On the other hand, the two "Type B" direct objects found in SOV clauses were not lengthy. They each contains two bunsetsu units. The frequency of the occurrence of OSV clauses in "Type A" (22.9%) is still higher than that in the data as a whole (14.1%). However, 77.1% (27/35) of the "Type A" direct objects appear in the SOV clauses and only 22.9% (8/35) appear in the OSV clauses. Therefore, we can say that OSV clauses tend to occur when the direct objects containing demonstratives refer to entities in the preceding discourse. We observed in the previous section the tendency suggested in Saeki (1960, 1975), that in clauses, long elements tend to precede other elements. Based on this analysis, we can assume that direct objects tend not to be preposed when the subjects of sentences are lengthy. Most of the direct object phrases containing demonstratives in the data consist of one or two bunsetsu units, such as sore o 'that A C C in 21 and sonna ree o 'that kind of example A C C in 22. In the following example 26, the direct object phrase kono jimusho o 'this office A C C precedes the subject phrase kyoowa-ha o shiensuru ooku no hitobito ga 'many people who support the Republican faction NOM,' which contains a noun-modifying clause. 82 (26) Kono / iimusho o /, kyoowa-ha o / shiensuru / ooku no / this office ACC republican-faction ACC support many GEN hitobito ga / otozureta ga, sono naka ni wa, wakaki hi people NOM visited but that among in TOP young days no aanesuto heminguuei no sugata mo atta. of Ernest Hemingway GEN figure also was-there 'Many people who supported the Republican faction visited this office, and among them, there was also the figure of Ernest Hemingway in his younger days.' (Igirisu-biiki: 88) The direct object and the subject in example 26 contain two and four bunsetsu units, respectively. Example 25 does not observe the tendency for long elements to precede others in clauses; however, it observes the second tendency for elements containing demonstratives to precede others. We saw in example 26 that"+ Demonstrative" direct objects containing a smaller number of bunsetsu units can precede "- Demonstrative" subjects containing a greater number of bunsetsu. On the contrary, there are cases in which "+ Demonstrative" direct objects containing a smaller number of bunsetsu units follow "- Demonstrative" subjects containing a larger number of bunsetsu units as shown in examples 27 and 28. (27) Kotoni yuushuuna / hitsujikai to / inu ga / kore o / yam especially excellent shepherd and dog NOM this ACC do toki nado, kenbutsunin ni tsutawaru nekki to koofun wa when EMPH spectator DAT convey heat and excitement TOP susamajii hodo da. terrific to the extent CPL 'The heat and excitement conveyed to spectators is terrific, especially when a skilled shepherd and his dog perform this.' (Igirisu-biiki: 111) (28) i<Kataribe> no / mukoogawa' to / daishita / rensai-kiji ga, / a clan of reciters GEN on the other side QT titled serial-article NOM konna / iookyoo o / hookokushita. such situation ACC reported 'The series of articles titled "on the other side of a clan of reciters" reported such a situation.' (Tenseijingo '97 aki Vol. 110: 84) 83 Examples 27 and 28 both contain "+ Demonstrative" direct objects; however, they do not precede subjects which are "- Demonstrative." Examples 29 and 30 below contain a clause with SOV word order which consists of subjects containing one bunsetsu unit and "+ Demonstrative" direct objects. (29) Shikashi, seefu wa / sono / yakusoku o / kutsugaeshita. but government TOP that promise ACC reversed 'However, the government reversed that promise.' (Igirisu-biiki: 83) (30) Shitagatte, taishikan ga / konoyoona / chansu o / ataete-kureru therefore embassy NOM such chance ACC provide-give koto wa, taihen, yuuigina koto da to watashi wa omotteiru. NMR TOP very significant thing CPL QT I TOP think 'Therefore, I think that fact that the embassy gives (us) such a chance is a very significant thing.' (Saigo no hanadokei: 37) The "+ Demonstrative" direct objects contain two bunsetsu units and the subjects contain one bunsetsu unit each in both 29 and 30. Therefore, the direct objects in 29 and 30 both have two structural characteristics which tend to place them in positions preceding other elements: they are longer and contain demonstratives. However, these direct objects do not precede the subjects. In the data, there are two clauses in which both the subjects and the direct objects are "+ Demonstrative." In both cases, the direct objects were not preposed. Examples 31 and 32 are clauses in which both the subjects and direct objects are "+ Demonstrative." (31) Kono budapesuto no hitobito ga, sore o shiranakatta wake this Budapest GEN people NOM it ACC did-not-know case wa nai no da. TOP NEG NMR CPL 'It cannot be the case that these people of Budapest did not know it.' (Uwanosora: 141) (32) Sono naka de sukunakaranu kazu no hitobito ga, that among in not-little number GEN people NOM shimoyama-san no kono hon o toriageteita. Shimoyama-Mr. GEN this book ACC were-mentioning 'Many people among them were mentioning this book of Mr. Shimoyama's.' (Uwanosora: 166) 84 It is thus possible to hypothesize that basic SOV word order occurs when both a subject and a direct object contain demonstratives. This requires further investigation but remains outside the scope of this thesis. This section examined the influence demonstratives in direct objects have on word order variation. As shown in Table 11, the frequency of occurrence of OSV clauses is greater among clauses containing "+ Demonstrative" direct objects than those in the data as a whole. However, the distribution of SOV and OSV clauses among clauses containing "+ Demonstrative" direct objects shows that about 70% (29/41) of "+ Demonstrative" direct objects do not appear in the sentence-initial position. As with the length of direct objects, although demonstratives contained in direct objects have some influence on word order, this does not on its own clearly identify the environment in which OSV clauses occur. 4.4 Idiomatic Expressions This section examines the third hypothesis of this thesis, which concerns direct objects forming idiomatic expressions together with verbs, as shown below: (33) Nihon-rettoo ga iki o_ nonda. Japanese-islands NOM breath ACC swallowed 'The Japanese islands were thrilled.' (Tenseijingo '97 aki Vol. 110: 15) (34) Soo to shitta kake no aite ga te o mawashita so QT learned betting GEN opponent NOM hand ACC sent no da. NMR CPL 'The opponent of the betting who had learned so arranged (things so that he would win).' (Tenseijingo '97 aki Vol. 110: 66) The bold faced iki 'breath' and nonda 'swallowed' together form an idiomatic expression iki o nonda 'thrilled' in 33, and te 'hand' and mawashita 'sent' form te o mawashita 'arranged' in 34 respectively. There are eleven such idiomatic expressions formed with a direct object and a verb in the data used in this thesis. Table 13 shows the distribution of SOV and OSV clauses and idiomatic 85 expressions in the data: SOV OSV Total Direct objects Idiomatic expressions 11 91.7% 1 8.3% 12 100% (4.1%) (2.3%) (3.9%) Others 256 88.6% 43 14.4% 299 100% (95.9%) (97.7%) (96.1%) Total 267 85.9% 44 14.1% 311 100% (100%) (100%) (100%) x2=0.35, df=l, NS (p>0.05) Table 13. Distribution of the direct objects forming idiomatic expressions All of the eleven direct objects forming idiomatic expressions in the SOV clauses contain one bunsetsu unit each as in examples 33 and 34. As shown in the above table, only one direct object forming an idiomatic expression appears in the clause-initial position, as the example below also illustrates: (35) Soo ii-nagara, watashi wa sooitta taishuu shoohi bunka no so say-while I TOP such popular consumption culture GEN kataboo o , kyuusokuni yutaka ni natta nihonjin to (take) part (in) ACC rapidly rich to became Japanese people and japan-manee ga katsuideiru koto ni omoi-itatta ga, Japanese-money NOM taking (part) in NMR to think-reached but sore wa iwazuni oita. it TOP without-saying left 'While saying so, I thought of the fact that Japanese people who have become rich rapidly and Japanese money have contributed to such popular consumption culture; however, I did not mention it.' (Itaria tosukaana no yuugana shokutaku: 75) The idiomatic expression kataboo o katsuideiru 'taking part in' often appears with a genitive 86 phrase modifying kataboo such as sooitta taishuu bunka no 'such popular consumption culture GEN' as in 35. On the other hand, the direct objects in the idiomatic expressions in 33 and 34 cannot appear with noun-modifying phrases.12 The idiomatic meaning 'thrilled' in 33 and 'arranged' in 34 would be lost, and the sentences would become unnatural. Furthermore, kataboo itself means 'one end of the pole of a palanquin' and katsuideiru means 'carrying.' Katsugu 'carry' is often used by itself literally meaning 'carry.' However, kataboo is rarely used meaning 'one end of the pole of a palanquin' since the Japanese no longer use palanquins as a method of transportation. Therefore, kataboo mostly appears in the idiomatic expression kataboo o katsugu 'take part in' as in example 35. This explains that example 35 sounds natural although the subject phrase kyuusokuni yutaka ni natta nihonjin to japan-manee ga 'Japanese people who have become rich rapidly and Japanese money NOM' intervenes between the direct object phrase kataboo o and the verb katsuideiru of the idiomatic expression kataboo o katsuideiru 'taking part in.' Compare the following examples, in which the direct objects in 33 and 34 appear in the sentence-initial position, to 35: (36) Tiki o nihon-rettoo ga nonda. breath ACC Japanese-islands NOM swallow 'The Japanese islands were thrilled.' (37) nTe o soo to shitta kake no aite ga hand ACC so QT learned betting GEN opponent NOM mawashita no da. sent NMR CPL 'The opponent of the betting who had learned so arranged (things so that he would win).' The direct object iki 'breath' in 36 and te 'hand' in 37 are frequently used by themselves, and as a result, the idiomatic meaning of 'thrilled' in iki o nonda and 'arranged' in te o mawashita appear very weak in examples 36 and 37. Consequently, 36 and 37 are unnatural sentences compared to 33 and 34, in which nothing intervenes between the direct objects and the verbs forming idiomatic 12Colher-Sanuki (personal communication) brought this point to my attention. 87 expressions. Therefore, the direct object component of an idiomatic expression made up of a direct object and a verb tends not to be preposed to the sentence-initial position in order to avoid an unnatural sentence. This section examined the influence of idiomatic expressions formed with a direct object and a verb on SOV and OSV word orders. The results in Table 13 support the hypothesis that direct objects in OSV clauses tend not to be a component of idiomatic expressions formed with a direct object and a verb. This thesis found only one such direct object among the forty-four OSV clauses in the data. 4.5 Emphasis This section examines the fourth hypothesis of this thesis, which concerns whether the direct object is interpreted as emphasized when it is preposed to the sentence-initial position. The preposed direct object is considered emphasized when it conveys focus by corresponding to the constituent that is wh-questioned in a question (Rochemont 1998). We have seen examples of OSV clauses in Section 4.2 (18, 19, and 20) that occur when the storyline centres around their direct objects. The following example also illustrates an emphasized direct object: 88 (38) [Furii sukuuru no sensee ga, futookoo no kodomo no fukki o free school GEN teacher NOM truant GEN children GEN return ACC mezashite gakkoo ni soodan ni iku. . . . Furii sukuuru no try school to consult purpose go free school GEN kyooryoku o kanarazushimo kangeeshinai.] cooperation ACC (not) necessarily not welcome Sonna ree o terebi ga hookokushiteita. such example ACC television NOM was-reporting [Kodomo ga iwayuru mondai koodoo o okosu. Jidoo child NOM so-called problematic behaviour ACC engage child soodan-jo de kaunseringu o ukeru yooni susumeru gakkoo consulting-place LOC counselling ACC receive so that recommend school ga am. Sore mo hitotsu no daijina hoohoo da. Shikashi. . .] NOM exist it also one GEN important method CPL however '[A teacher from a free school goes to consult with the regular school trying to return a truant student to the regular school. . . . The regular school does not necessarily welcome the cooperation of the free school.] A television program reported such an example. [A child engages in so-called problematic behaviour. There are schools which recommend the child receive counselling at the child consulting centre. That is also one important method. However . . .]' (Tenseijingo '97 aki Vol. 110: 78) The discourse preceding the OSV clause describes sonna ree 'such an example' of problems with truant students in Japan, and the difficulties encountered by independent schools when trying to communicate with regular schools. The discourse following this OSV clause further illustrates the problems with truant students. Therefore, the main information conveyed in the OSV clause, sonna ree o, terebi ga hookokushiteita 'a television program reported such an example,' in 38 is that sonna ree 'such an example' is reported, not that terebi 'a television program' reported something. When the OSV clause sonna ree o, terebi ga hookokushiteita 'a television program reported such an example' in 38 is expressed in SOV word order, the direct object sonna ree 'such an example' is not interpreted as emphasized as shown in 39: (39) Terebi ga sonna ree o hookokushiteita. television NOM such example ACC was-reporting 'A television program reported such an example.' 89 The interpretation of 39 is neutral compared to that of the OSV clause in 38. The same difference in interpretation between OSV and SOV clauses is also observed between the OSV clause in 40 and the SOV clause in 41: (40) Katsute furankii sakai ga netsuenshite. koohyoo once Franky Sakai NOM give-an-ardent-performance favourable-comment o hakushita 'watashi wa kai ni naritai' o ACC received I TOP shellfish into want-to-become ACC tokoro jooji ga enjite, futatabi 'watashi wa kai ni Tokoro George NOM perform again I TOP shellfish into naritai' ga terebi de hoosoosareta. want-to-become NOM television on was-broadcasted 'George Tokoro performed "I want to become a shellfish," of which Franky Sakai once gave an ardent performance and received favourable comments, and "I want to become a shellfish' was broadcast on television again." (Saigo no hanadokei: 146) (41) Tokoro jooji ga katsute furankii sakai ga netsuenshite. Tokoro George NOM once Franky Sakai NOM give-an-ardent-performance koohyoo o hakushita 'watashi wa kai ni favourable-comment ACC received I TOP shellfish into naritai' o enjite, futatabi 'watashi wa kai ni want-to-become ACC perform again I TOP shellfish into naritai' ga terebi de hoosoosareta. want-to-become NOM television on was-broadcasted 'George Tokoro performed "I want to become a shellfish," of which Franky Sakai once gave an ardent performance and received favourable comments, and "I want to become a shellfish" was broadcast on television again.' The OSV clause in example 40 appears at the very beginning of a section entitled wasurerarenu taiken 'Unforgettable Experience,' and is followed by a brief description of the story of the drama called watashi wa kai ni naritai T want to become a shellfish,' which is about a war criminal. This section continues to describe how the critics and audience were not able to relate themselves to this story, and thus did not think highly of this drama despite it being well-made. The author of this essay touches on the actor who played the main character in the drama tokoro jooji 'George Tokoro,' which is the subject, but the main information of this OSV clause is the drama, not the 90 actor. On the other hand, the direct object katsute furankii sakai ga netsuenshite, koohyoo o hakushita 'watashi wa kai ni naritai' "T want to become a shellfish," of which Franky Sakai once gave an ardent performance and received favourable comments' in 41 is not as strongly asserted as it is in 40. Table 14 below illustrates the distribution of direct objects interpreted as emphasized in the collected OSV clauses : Emphasized direct object Others Total OSV clauses 40 90.9% 4 9.1% 44 100% x 2=29.46, df=l, p<0.05 Table 14. Distribution of the emphasized direct objects in the O S V clauses Forty direct objects (90.9%) of forty-four OSV clauses are interpreted as emphasized in the collected data; therefore, we can say that the direct object in an OSV clause is emphasized. One of the forty emphasized direct objects is a wh-phrase as shown in example 42 below: (42) Nani o kare wa tsutaetakatta no ka. what ACC he TOP wanted-to-convey (a message) NMR Q '(I wonder) what message he wanted to convey.' (Tenseijingo '97 aki Vol. 110: 75) The preposed direct object nani 'what' is neither lengthy nor contains a demonstrative. Furthermore, the preposing of wh-phrase nani o 'what A C C in 42 is optional in Japanese, thus example 42 can be expressed without preposing the wh-phrase nani o 'what A C C as shown below: (43) Kare wa nani o tsutaetakatta no ka. he TOP what ACC wanted-to-convey (a message) NMR Q '(I wonder) what message he wanted to convey.' In contrast, English requires the preposing of wh-phrases in questions and embedded questions as 91 shown in the translations of 42 and 43. Kuno (1972: 289) claims that the difference between a question with a preposed wh-phrase as 42, and one without, as 43, is that in the former, what follows the wh-phrase "acquires the subordinate clause-like nature." In other words, kare wa tsutaetakatta 'he wanted to convey,' which follows the preposed wh-phrase nani o 'what ACC,' is secondary information, and the preposed wh-phrase nani o 'what A C C is the main information in 42. Therefore, the interpretation of sentences 42 and 43 are slightly different in that the wh-phrase in 42 is interpreted as emphasized. Four direct objects of forty-four OSV clauses in the data are not interpreted as emphasized, as shown in Table 14. The following is an example of such a direct object. (44) [Kimu to riza wa, pari no jibun-tachi no apaato o, Kim and Liza TOP Paris GEN self-plural GEN apartment ACC kyoowa-ha no renraku-jimusho toshite kaihooshita. ] republican-faction GEN correspondence-office as opened Kono iimusho o. kyoowa-ha o shiensuru ooku no this office ACC republican-faction ACC support many GEN hitobito ga otozureta ga, sono naka ni wa, wakaki hi no people NOM visited but that among in TOP young days of aanesuto heminguuei no sugata mo atta. Ernest Hemingway GEN figure also was-there '[Kim and Liza opened their apartment in Paris as a correspondence office of the Republican faction.] Many people who supported the Republican faction visited this office, and among them,there was also the figure of Ernest Hemingway in his younger days.' {Igirisu-biiki: 88) The discourse preceding 44 is about how Kim came to support the Soviet Union's Republican faction as well as the Republican faction's fight in Europe. The OSV clause Kono iimusho o, kyoowa-ha o shiensuru ooku no hitobito ga otozureta 'Many people who supported the Republican faction visited this office' is followed by further information on what kind of people visited this office. Following 44, further description is given of Kim's support for the Republican faction and his actions during the Second World War. Since neither what follows the OSV clause Kono jimusho o, kyoowa-ha o shiensuru ooku no hitobito ga otozureta 'Many people who 92 supported the Republican faction visited this office' in 44, nor the discourse following 44, is about the direct object kono jimusho 'this office,' the OSV clause is not interpreted as being about the direct object. In other words, this direct object is not interpreted as emphasized although it is preposed to the clause-initial position. Notice that the direct object kono jimusho 'this office' contains the demonstrative kono 'this.' The OSV clause in 45 below also contains a direct object containing a demonstrative and is not interpreted as emphasized. (45) [Onaji doits u-minzoku da shi, fasshon no hairi-kata ni mo, same German-nation CPL furthermore fashion GEN enter-way in also jisa wa nasasoo da.] Daga kono naka no ooku time difference TOP seem-not-to-exist CPL however this among GEN many no hitotachi ga shitsugyoosha de sono seekatsu o nishi-gawa GEN people NOM unemployed CPL their life ACC west-side shimin ga sasaeteiru. citizens NOM supporting '[(They are) the same Germans and furthermore, there seems to be no time difference in the way fashion is accepted as well.] However, many of them are unemployed, and the citizens in the West Germany are supporting their lives.' (Uwanosora: 149) The preceding discourse of 45 is about difficulties people in the eastern and western parts of Germany have been experiencing because of the differences in quality of life since the destruction of the Berlin wall. In this context, the direct object sono seekatsu 'their life' is not interpreted as emphasized. We have seen in Section 4.3 that the direct object containing a demonstrative tends to be preposed to the sentence-initial position. The preposing of the direct object kono jimusho 'this office' in 44 and sono seekatsu 'their lives' in 45 may be due to the demonstratives, not emphasis. The other two of four direct objects, which are not interpreted as emphasized, do not contain demonstratives. Below is the first of two examples (46 and 47) containing such a direct object. 93 (46) Dosekiryuu de giseesha ga deta kagoshima-ken no_ landslide because-of victim NOM be caused Kagoshima-prefecture GEN genchi o kamei shizuka kensetsu-soo ga shisatsushita. site . ACC Kamei Shizuka Construction-Minister ACC visited 'Construction Minister Shizuka Kamei visited the site in Kagoshima Prefecture where many people had been killed in a landslide.' {Tenseijingo '97 aki Vol. 110: 24) Example 46 appears following a clause wadai no daini 'the second topic' indicating the beginning of the second half of the essay, which talks about two topics from the week before regarding two politicians. The discourse following 46 describes what the minister did during his visit to the site, so the direct object in 46 is not interpreted as emphasized in this context. Notice that the preposed direct object contains a clause. Example 46 contains a clause which provides "a characterization or description of a New head NP referent, not previously known to the hearer" (Fox and Thompson 1990: 301). Therefore, the preposing of the direct object in 46 does not emphasize this direct object, but it provides information which will make example 46 more relevant to the reader. Section 4.6 further discusses the function of relative clauses. The following is the other example of an OSV clause containing a direct object that is not emphasized. (47) Nimokakawarazu "saibansho ga ninteeshita hanzai-iiiitsu o nevertheless court NOM established criminal-facts ACC watashi wa mitomenai" nado to kyoobenshita. I TOP do-not-acknowledge EMPH QT asserted 'Nevertheless, he asserted, "I do not acknowledge the criminal facts established by the court.'" (Tenseijingo '97 aki Vol. 110: 133) As shown above, the OSV clause in 47 quotes Congressman Koko Sato's comment on the court's decision, which is described in the preceding discourse. Since the context in which this OSV clause is uttered is unknown, it is impossible to determine whether the preposed direct object saibansho ga ninteeshita hanzai-jijitsu 'the criminal facts established by the court' is emphasized or not. This section examined the relationship between the preposing of the direct object and its interpretation as emphasized or unemphasized. The results in Table 14 indicate that direct objects 94 in OSV clauses tend to be interpreted as emphasized. The examples provided also show that the preposed direct objects interpreted as emphasized lose this interpretation when they appear in SOV word order. 4.6 Information Status of Direct Objects This section examines the fifth hypothesis of this thesis, which concerns the relationship between direct objects in OSV clauses and the preceding discourse. Following the taxonomy presented in Chapter 2, which is repeated below, the information status of direct objects is examined.13 Information status of NPs Discourse-new/ Hearer-new Unanchored Anchored Discourse-new/ Hearer-old Unused Inferrable Discourse-old/ Hearer-old Situationally Textually Evoked Evoked Figure 4. Information status of discourse entities: taxonomy of the thesis In the following section, I demonstrate examples of NPs, from the data, of each information status. When the examples include parts from the preceding discourse, these parts are indicated with square brackets. "Unanchored" and "Anchored" NPs are those which are newly introduced to the discourse and the hearer.14 Sentences 48 and 49 are examples of "Unanchored" and "Anchored" direct objects, respectively. 1 3See Table 3 or Appendix B for the definitions of each information status. 1 4 A 1 though this thesis analyzes written discourse, I use 'hearer' instead of 'reader' in order to be consistent with the term "Hearer-old/new." 95 (48) "Unanchored" direct object Hayashi-san wa vuuforia toyuu kotoba o gozonjidesu ka. Hayashi-Mr. TOP euphoria COMP word ACC know Q 'Mr. Hayashi, do you know the word "euphoria"?' (Igirisu-biiki: 16) (49) "Anchored" direct object ['Ame to muchi' toyuu kotoba o omoidashi-nagara, sakuyuu candy and whip COMP word ACC remember-while last evening moyoosareta seefu-shusai Okinawa fukki 25-shuunen kinen held government-sponsored Okinawa return 25-year anniversary shikiten' no terebi-chuukee o mita. ] ceremony GEN television-broadcast ACC watched Endan ni tatta hashimoto-shushoo wa, keezai-men nado de_ podium on stood Hashimoto-prime minister TOP economy-respect and others in Okinawa o yuuguusuru itsutsu no shisaku o yakusokushita. Okinawa ACC treat-well five GEN policies ACC promised '[(I) watched television coverage of the government-hosted "ceremonies to commemorate the 25th anniversary of Okinawa's return," which were held last evening, thinking about the "carrot and stick" anecdote.] Prime Minister Hashimoto, who stood on the podium, promised five policies which would help the Okinawa prefecture in respect to economics and other elements.' {Tenseijingo '97 fuyu Vol. 110: 96) The direct objects yuuforia toyuu kotoba 'the word euphoria' in 48 and keezai-men nado de Okinawa o yuuguusuru itsutsu no shisaku 'five policies which would help the Okinawa prefecture in respect to economics and other elements' in 49 are both "Discourse-new" because they have not been mentioned in the preceding discourse and "Hearer-new" because they are considered not to be known to the hearer. The two direct objects in 48 and 49 are different in that the direct object in 49 contains an NP, an 'anchor,' which the direct object in 48 does not contain, linking this direct object with other NPs in the preceding discourse. The bold faced NP Okinawa 'the Okinawa prefecture' of the direct object Okinawa o yuuguusuru itsutsu no shisaku 'five policies which would help the Okinawa prefecture in respect to economics and other elements' in 49, which has already been mentioned in the sentence appearing in the preceding discourse, functions as an 'anchor'; thus, the direct object in 49 is linked to the preceding discourse. 96 "Unused" NPs are those which are "Discourse-new" and "Hearer-old." The direct object in 50 is an example of an "Unused" NP. (50) "Unused" direct object Satoo kookoo-shi ga soomuchoo-chookan o jininshita. Sato Koko-Mr. NOM secretary-general ACC resigned 'Mr. Sato Koko resigned his position as a secretary-general.' {Tenseijingo '97 aki Vol. 110: 170) Sentence 50 is the first sentence of a short article; therefore, the direct object is "Discourse-new." Although the direct object soomuchoo-chookan 'secretary-general' is introduced to the hearer for the first time, this noun is considered to already be known to the hearer; thus, it is an "Unused" NP. 1 5 The direct object in 51 is an example of an "Inferrable" NP. l 5 What is "Unused" may vary depending on the knowledge of a given individual. For example, we expect most adult speakers to know the term soomuchoo-chookan 'secretary-general' in example 50 but we do not expect children to know it. For this thesis, I categorize NPs that are "Discourse-new" and considered to already be known to Japanese speakers in their late 20's and early 30's as "Unused." 97 (51) "Inferrable" direct object [Ashimoto o kaichuudentoo de terashi-nagara, natsu to wa close-to-my-feet ACC flashlight with light-while summer QT TOP ie yama no tsumetai yaki ni hada o sarashi-nagara say-even mountain GEN cold night-air to skin ACC expose-while kuruma no tokoro e modotta ga, nando tameshite-mo car GEN place to returned but no-matter-how-many-times try-even kuruma no enjin ga kakaranai. Nemuke mo ippen-ni car GEN engine NOM doesn't-start sleepiness EMPH at-once fukitondeshimatta. ] blew-away Juriaana ga, heddoraito o keshi-wasureteita no da. Juliana NOM headlight ACC forgot-to-turn-off NMR CPL '[Lighting the area close to my feet with a flashlight, we returned to the place where we had parked our car, exposing our skin to the mountain night air, which was cold even though it was summer. However, no matter how many times we tried, the engine of the car would not start. Our sleepiness disappeared at once.] Juliana had forgotten to turn off the headlights of our car.' (Itaria tosukaana no yuugana shokutaku: 122) The direct object heddoraito 'headlight' in 51 is "Inferrable" from the bold faced NP kuruma 'car' which appears in the preceding discourse. There are two types of "Discourse-old" and "Hearer-old" NPs: "Situationally Evoked" and "Textually Evoked." In the following sentence 52, the subject is "Situationally Evoked": (52) Watashi wa seesho ni kakareta iesu no kiseki o omoidashita. I TOP Bible in written Jesus GEN miracle ACC recalled T recalled the miracle of Jesus written in the Bible.' (Saigo no hanadokei: 169) The subject watashi T appeared in the text for the first time. This watashi T refers to the author, who exists in an extratextual context. Therefore, it is "Situationally Evoked." Sentence 53 is an example of a "Textually Evoked" NP: 98 (53) "Textually Evoked" direct object [Popai no koibito oriibu oiru sokkuri. ] Popeye GEN sweetheart Olive Oil look-exactly-like Koigataki no buruuto ga, oriibu o saraootosuru. rival in love GEN Brutus NOM Olive ACC try-to-kidnap '[(She) looks exactly like Popeye's sweetheart Olive Oil . . . . ] Brutus, Popeye's rival in love, tries to kidnap Olive.' {Tenseijingo '97 aki Vol. 110: 34) The direct object oriibu 'Olive' in the SOV clause in 53 is mentioned for the second time. It has been mentioned in the preceding discourse as Popai no koibito oriibu oiru 'Popeye's sweetheart Olive Oil' as shown in 53. Table 15 shows the distribution of direct objects in six information statuses in the collected SOV and OSV clauses. 99 SOV OSV Total Discourse-new/Hearer-new Unanchored 70 95.9% (26.2%) 3 4.1% (6.8%) 73 100% (23.5%) Anchored 68 70.1% (25.4%) 29 29.9% (65.9%) 97 100% (31.2%) Discourse-new/Hearer-old Unused 1 50% (0.4%) 1 50% (2.3%) •• 2 100% (0.6%) Inferrable 60 95.2% (22.5%) 3 4.8% (6.8%) 63 100% (20.3%) Discourse-old/Hearer-old Situationally Evoked 0 0% (0%) 0 0% (0%) 0 0% (0%) Textually Evoked 68 89.5% (25.4%) 8 10.5% (18.2%) 76 100% (24.4%) Total 267 85.9% (100%) 44 14.1% (100%) 311 100% (100%) x2=11.07, df=5, p<0.05 Table 15. The information status of direct objects I did not find "Situationally Evoked" direct objects in the data used in this thesis. Therefore, the data of this thesis are unable to verify the hypothesis that the direct object in OSV clauses tends to be closely related to NPs in the nonlinguistic context. The SOV word order occurs more frequently than the OSV word order in all information statuses except for the "Unused" one; the data of this thesis contain only two "Unused" direct objects in total, one in an SOV clause and 100 the other in an OSV clause. SOV clauses contain about the same number of "Unanchored," "Anchored," "Inferrable," and "Textually Evoked" direct objects (26.2%, 25.4%, 22.5%, and 25.4%, respectively). On the other hand, about two-thirds of the direct objects in OSV clauses are "Anchored" (65.9%). This may be due to the characteristics of "Anchored" NPs. As shown in example 49 repeated below, "Anchored" NPs contain an 'anchor' NP which links the "Anchored" NP to other discourse entities; therefore, they tend to be long. As presented in section 4.2, the average number of bunsetsu units in direct objects in the data is 3.07. On the other hand, the average number of bunsetsu units in the "Anchored" direct objects is 5.35, which is approximately 1.7 times greater than the average number in the data as a whole. Furthermore, twenty-four out of the twenty-nine (82.8%) of the "Anchored" direct objects in the OSV clauses are "+ Clause." SOV clauses contain forty "+ Clause" "Anchored" direct objects, as in the SOV clause in 49 repeated below, which is 58.8% of its sixty-eight "Anchored" direct objects. Examples 54 and 55 illustrate OSV clauses containing "Anchored" "+ Clause" direct objects. (49) Endan ni tatta Hashimoto-shushoo wa, keez.ai-men nado de podium on stood Hashimoto-prime minister TOP economy-respect and others in Okinawa o yuuguusuru itsutsu no shisaku o yakusokushita. Okinawa ACC treat-well five GEN policies ACC promised 'Prime Minister Hashimoto, who stood on the podium, promised five policies which would help Okinawa prefecture with respect to economics and other elements.' (Tenseijingo '97 fuyu Vol. 110: 96) (54) Ikura kawaii ko ? dakara to itte sonnani tooku made tabi o how dear child because QT say such far to travel ACC saseyoo toyuu oya no kokoro o , watashi wa . let-will COMP parent GEN thought ACC I TOP hakari-kanete-shimatta measure-cannot-EMPH T couldn't understand my parents' way of thinking, which would allow me travel so far even though I was such a dear child.' (Yotsuba no essei: 136) 101 (55) Dairi shomee nado chiii-ra ga. representative signing such as prefectural governor-plural NOM katagawarishite-kita shigoto o kuni ga tantoosuru. take over-have been work ACC national government NOM take charge 'The national government takes charge of works, which the prefectural governors had been taking over, such as representative signing.' (Tenseijingo '97 aki Vol. 110: 127) Fox and Thompson (1990: 301) point out that one of the functions of the relative clause is to "provide a characterization or description of a "New" head NP referent, not previously known to the hearer."16 We can observe this function of the relative clause in examples 49, 54, and 55. The head NPs, shisaku 'policies,' kokoro 'thought,' and shigoto 'work' are "Discourse-new," and the relative clauses, which contain 'anchor' NPs, provide more information about these head NPs in the above examples. Fox and Thompson claim that "referents are presented to so as to be RELEVANT for listeners at the point where they are introduced" in effective communication, and "grounding" is a primary way in which speakers make an NP relevant. They define the "grounding" function as "to locate an NP's referent in conversational space by relating it to a referent whose relevance is clear, that is, to a "Given" referent in the immediate context" (1990: 1 *7 300). They further claim that 'anchoring' the relative clause is one way to "ground" an NP; an 'anchored' relative clause makes the head NP relevant by relating it to the preceding discourse. Therefore, according to Fox and Thompson, although direct objects in 49, 54, and 55 are new to the discourse and the hearer, they are made relevant due to the 'anchored' relative clauses. Having preposed direct objects convey "Discourse-new" information relevant to the discourse is possible due to the structure of the Japanese relative clause. Collier-Sanuki (1993) demonstrates the difference between the structures of relative clauses with object heads in Japanese 1 6 Fox and Thompson (1990: 299) follow Chafe (1980, 1987) and define "New" information as "a referent introduced into the discourse, presumed not to be in the hearer's focal consciousness." 1 7 Fox and Thompson (1990: 300) follow Chafe (1980, 1987) and define "Given" as "a referent presumed to be in the hearer's focal consciousness." 1 8 For other kinds of "grounding," see Fox and Thompson (1990). 102 and English as follows: (56) a. structure of Japanese relative clause with object head: (Subject) [RCJObjectNP Verb, b. structure of English relative clause with object head: Subject V ObjecfNP[RC]. (Collier-Sanuki 1993:65) As shown in 56a, the relative clause precedes the head object NP in Japanese; thus, the head object NP is grounded by the subject and the relative clause in the canonical word order. On the other hand, relative clauses follow the head object NP in English; therefore, the object head NP is grounded by the subject and the verb, but not by the relative clause, as shown in 56b. The object head NP in Japanese is grounded by the relative clause when the object NP is preposed to the clause-initial position as shown in 57: (57) structure of Japanese relative clause with object head in OSV word order: [RCjObjectNP (Subject) Verb. Since the object head NP will be grounded by the relative clause, the direct object conveying "Discourse-new" information can be made relevant to the discourse by means of the relative clause even when it is preposed. There are five "Anchored" "- Clause" direct objects in the OSV clauses in the data. Although these direct objects do not contain clauses providing a grounding function, their relevance is made clear by 'anchor' NPs preceding the direct object NPs. Example 58 contains such "Anchored" direct objects in OSV clauses. 1 9 For further discussion of the differences between Japanese and English relative clauses, see Collier-Sanuki (1993). 103 (58) Soshite kono geemu no torikku wa pieeru burujyuu no yuu toori, and this game GEN trick TOP Pierre Brujeau GEN say as geemu no ruuru o daitasuu no haisha ga sasaeteiru game GEN rule ACC large majority of defeated NOM support toyuu koto da. COMP NMR CPL 'And the trick of this game is, as Pierre Brujeau says, that the large majority of defeated people are supporting the rule of the game.' (IJwanosora: 173) The 'anchor' NP geemu 'game' precedes the direct object NP ruuru 'rule' in 58. Therefore, this direct object NP is relevant at the point at which it is introduced. "Unanchored" NPs, which are also "Discourse-new/Hearer-new," are not linked to the discourse since these NPs do not contain 'anchors' linking them to the discourse. However, "Unanchored" NPs can be linked to the hearer's knowledge. There are two such direct objects in the data of this thesis. Example 40 repeated below and example 59 illustrate the OSV clauses with "Unanchored" direct objects: (40) Katsute furankii sakai ga netsuenshde. koohyoo once Franky Sakai NOM give-an-ardent-performance favourable-comment o hakushita "watashi wa kai ni naritai" o ACC received I TOP shellfish into want-to-become ACC tokoro jooji ga enjite, futatabi "watashi wa kai ni Tokoro George NOM perform again I TOP shellfish into naritai" ga terebi de hoosoosareta. want-to-become NOM television on was-broadcasted 'George Tokoro performed "I want to become a shellfish," of which Franky Sakai once gave an ardent performance and received favourable comments, and "I want to become a shellfish" was broadcast on television again.' (Saigo no hanadokei: 146) (59) Ichioku-en ijoo no okane o , futsuu no hito one hundred million-yen more than of money ACC ordinary GEN people ga soozoo-dekinai no to sore wa niteiru. NOM imagine-cannot NMR to it TOP similar 'It is similar to the [fact that] ordinary people cannot imagine an amount of money that is greater than one hundred million yen.' (Yotsuba no essei: 12) The direct objects in 40 and 59 both contain NPs that convey "Unused" information to the hearer: 104 furankii sakai 'Franky Sakai' and ichioku-en 'one hundred million yen' respectively . These direct objects are linked to the hearer's knowledge by these "Unused" NPs, thus, the direct objects in 40 and 59 are relevant to the hearer when they are introduced to the discourse. One of the three "Unanchored" direct objects in OSV clauses in the data, example 42 repeated below, does not contain "Unused" information. Instead the direct object is a question word. (42) Nani o kare wa tsutaetakatta no ka. what ACC he TOP wanted-to-convey (a message) NMR Q '(I wonder) what message he wanted to convey.' {Tenseijingo '97 aki Vol. 110: 75) The direct object nani 'what' in 42 appears in the sentence-initial position; however, it is neither lengthy nor contains a demonstrative. Section 4.5 discusses this example. There are two OSV clauses that appear at the beginning of discourses, as previously shown by example 40 and example 46 repeated below: (46) Dosekiryuu de giseesha ga deta kagoshima-ken no_ landslide because-of victim NOM be caused Kagoshima-prefecture GEN genchi o kamei shizuka kensetsu-soo ga shisatsushita. site ACC Kamei Shizuka Construction-Minister NOM visited 'Construction Minister Shizuka Kamei visited the site in Kagoshima Prefecture where many people had been killed in a landslide.' (Tenseijingo '97 aki Vol. 110: 24) Notice that the direct objects in 40 and 46 both contain clauses providing "a characterization or description of a New head NP referent, not previously known to the hearer" (Fox and Thompson 1990: 301). Examples 40 and 46 show that "Discourse-new" direct objects containing clauses can appear not only in the sentence-initial position, but also at the beginning of discourses. The data do not contain "Discourse-new" direct objects without clauses appearing at the beginning of discourses. All the "Textually Evoked" direct objects in OSV clauses are "-(-Demonstrative." In the previous section, we saw two types of "+ Demonstrative" NPs: NPs containing demonstratives which themselves refer to an entity in the preceding discourse or demonstratives which themselves are nouns (Type A), and NPs containing demonstratives which themselves do not refer to an entity in the preceding discourse (Type B). There are twenty-seven "Type A" and two "Type B" "+ 105 Demonstrative" direct objects in the SOV clauses, and eight "Type A" and four "Type B" direct objects in the OSV clauses. "Type A + Demonstrative" direct objects are "Discourse-old" since they refer to entities in the preceding discourse. An example of a "Type A + Demonstrative" direct object, 21, is repeated below. (21) Senpai. kangofu ga sore o hagemashiteiru. senior nurse NOM it ACC encouraging 'A senior nurse encourages it (=a new nurse crying).' (Saigo no hanadokei: 131) All of the six "Type B + Demonstrative" direct objects found in the data are "Discourse-new" as shown in 23 and 24 repeated below. (23) Nishi-berurin ni nijuu-nen sundeiru doitsujin no yuujin wa West Berlin LOC twenty-year live German GEN friend TOP sono sonzai o shirazu, infomeeshon sentaa ni itte kiite-mo its existence ACC don't-know information centre to go-and ask-even yooryoo-o-enai. be-off-the-point 'A German friend, who has been living in West Berlin for twenty years, did not know of its existence, and even when I went and asked at the information centre, their information was unclear.' (Uwanosora: 126) (24) Sonna shakaiteki jookyoo o mushishite ichi-oku o such social circumstance ACC ignore one-hundred million ACC koeru kane o hirooen ni kakeru shinkee o watashi wa utagau exceed money ACC reception on spend nerve ACC I TOP mistrust kara dearu. because CPL 'It is because I mistrust the nerve that ignores such social circumstances and spends an amount of money exceeding one hundred million yen on a wedding reception.' (Saigo no hanadokei: 31) The direct objects in 23 and 24 contain demonstratives which link these direct objects to entities in the preceding discourse; however, the direct objects themselves are new to the discourse. Therefore, they are "Anchored" "Discourse-new" direct objects. Among the sixty-nine "Textually Evoked" direct objects in the collected SOV clauses, twenty-seven are "Type A + Demonstrative" and forty-two are "- Demonstrative" direct objects. 106 As for the OSV clauses, all of the eight "Textually Evoked" direct objects are "Type A + Demonstrative" direct objects. Therefore, except for the eight (18.2%) direct objects containing demonstratives which refer to entities in the preceding discourse, all the direct objects in the OSV clauses are "Discourse-new." This section examined the hypothesis that direct objects in OSV clauses are closely related to entities in the preceding discourse or entities in the nonlinguistic context. The data shows that it is not necessary for direct objects to be "Discourse-old" to precede subjects in clauses. The direct objects in OSV clauses, from the data, contain only 18.2% (=8/44) of the direct objects previously mentioned, and 81.8% (=36/44) were newly introduced direct objects. The data also show that most of the newly introduced preposed direct objects are linked to the discourse. This thesis did not find enough data to verify the hypothesis that the preposed direct objects are closely related to entities in the nonlinguistic context. 4.7 Summary This chapter analyzed the collected 311 SOV and OSV clauses to verify the five hypotheses of this thesis as repeated below: (1) The direct objects in OSV clauses a. tend to be lengthy compared to those in SOV clauses; b. tend to contain demonstratives unlike those in SOV clauses; c. tend not to form an idiomatic expression together with a verb; d. tend to be an emphasized NP; and e. tend to be closely related to entities in the preceding discourse or entities in the nonlinguistic context. The results found in Sections 4.2 to 4.6 support the above five hypotheses, except for hypothesis lb and the second part of hypothesis le. As per lb, this thesis found that 10.9% (29/267) of the SOV clauses and 27.3% (12/44) of the OSV clauses had direct objects containing demonstratives. Therefore, although the OSV clauses had higher percentage of direct objects containing demonstratives, the data did not verify that direct objects in SOV clauses tend not to contain 107 demonstratives. As per the second part of le, this thesis did not find enough data to verify this hypothesis that preposed direct objects tend to be closely related to entities in the nonlinguistic context. The following summarizes the findings of the quantitative analyses from each section: (60) The preposed direct objects in OSV clauses a. tend to be lengthy (Section 4.2, Tables 8 and 9); b. tend to contain clauses (Section 4.2, Table 10); c. tend to contain demonstratives compared to the direct objects in SOV clauses (Section 4.3, Table 11); d. tend not to be a component of idiomatic expressions formed with a direct object and a verb (Section 4.4, Table 13); e. tend to be interpreted as emphasized (Section 4.5, Table 14); and f. tend to convey information linked to the preceding discourse (Section 4.6, Table 15); This thesis found that some preposed direct objects in the data exhibit two or more of the above characteristics and some only one. On the other hand, some direct objects are not preposed although they exhibit one or more of the above characteristics in 60a to 60f. This indicates that none of the above characteristics fully explains the OSV word order on its own, so what motivates the preposing of the direct objects varies depending on what the context demands for each case. 108 Chapter Five Conclusion 5.1 Summary This thesis examined the characteristics of preposed direct objects in order to understand the difference between basic SOV word order and its OSV variation. Chapter 2 reviewed previous studies on preposing from two perspectives, syntactic analyses and pragmatic analyses. The review of syntactic analyses showed that the direct object in an OSV clause is preposed to the clause-initial position from its canonical position following the subject. The review of pragmatic analyses showed that preposed constituents have certain structural characteristics and convey certain information. Chapter 2 then presented the hypotheses of this thesis, which attempted to verify whether the previous claims made on preposing in general apply to OSV word order. Following the method described in Chapter 3, Chapter 4 proceeded to analyze the collected SOV and OSV clauses. The analyses demonstrated that the direct objects in OSV clauses a) tend to be lengthy; b) tend to contain clauses; c) tend to contain demonstratives; d) tend not to be components of idiomatic expressions; e) tend to convey information linked to the preceding discourse; f) tend to convey information linked to the hearer's knowledge; and g) tend to be emphasized. . The characteristics found in the direct objects in OSV clauses provide explanations for the preposing of direct objects in Japanese. The lengthy direct objects precede the shorter subjects in order to avoid intervening between the subjects and the verbs, thereby creating clauses easier to understand. The preposing of direct objects containing demonstratives and of those conveying information linked to their preceding discourse or hearer's knowledge make the OSV clauses relevant. Also, direct objects appear in the sentence-initial position in order to be interpreted as emphasized. This thesis found that though OSV word order does not occur frequently (14.1% (44/311) in the data of this thesis), it certainly has an important role in effective communication. 109 5.2 Limitations This thesis is limited in several ways due to its methodology. First, the findings of this thesis are limited to SOV and OSV word orders in written Japanese, since the data analyzed were collected from written materials. Previous studies by Saeki (1960, 1975) and Miyajima (1964) also analyze word order variations in written Japanese. As written language is more organized and integrated than spoken (Clancy 1988), this thesis was able to observe many OSV clauses with lengthy direct objects. On the other hand, spoken language is more spontaneous, thus OSV clauses found in spoken Japanese may not be due to the length of the preposed direct objects. Moreover, speakers can indicate emphasis by vocally stressing phrases they wish to emphasize, but writers cannot. Thus, understanding of OSV word order could be advanced through the analysis of spoken data. It is yet to be examined, for example, whether speakers employ preposing to indicate emphasis as writers do. Second, the findings are limited to certain types of SOV and OSV clauses. The data exclude clauses containing a) constituents other than subjects, direct objects, and verbs, such as indirect objects and adverbs; b) direct objects marked with the topic/contrast marker wa; c) direct objects marked with the nominative case marker ga; d) subjects and direct objects marked with kakari-joshi; and e) a subject marked with wa which contrasts strongly to another wa-marked subject in the same sentence. By excluding such clauses, this thesis was able to focus strictly on the differences between SOV and OSV clauses based on the basic functions of subjects and direct objects. However, it was unable to provide an explanation for the preposing of direct objects in general. As demonstrated in Chapter 3, the preposing of direct objects does not normally occur when direct objects are marked with ga (as in (c)) or when the clause contains a wa-marked subject that contrasts strongly to another wa-marked subject in the same sentence (as in (e)). On the other hand, preposing direct objects is possible in environment (a) and (d) above. Moreover, wa-marked direct objects (as in (b)) are preposed on separate grounds, since sentence topics tend to precede other constituents (Kuno 1972). Therefore, analyzing the preposing of direct objects in various environments, and comparing characteristics of preposed direct objects and topicalized 110 direct objects could contribute to knowledge of OSV word order in Japanese. 5.3 Suggestions Further Study There are some areas which require further investigation. First, the results of the data analyses show that the direct objects containing demonstratives tend to be preposed. However, this thesis did not have enough data to verify whether such direct objects are preposed, even when the subjects of the same clauses contain demonstratives. There were two clauses in which both the subjects and the direct objects contained demonstratives, and they both had SOV word order. Second, empirical studies of sentence processing may provide further evidence for functions of preposing. Studies such as that of Frazier and Rayner (1988), which measure the time taken to read sentences containing constituents appearing in different orders, will be useful for assessing whether preposing facilitates easier processing of clauses. Studies like Tomlin's (1995), which examines focal attention on visually marked information, may be able to verify whether preposed direct objects are indeed emphasized. Finally, analyses of the preposing of constituents other than direct objects should enable us to understand functions of preposing in Japanese in general. As this thesis revealed, though the preposing of direct objects is commonly considered optional, it is in fact motivated by reasons which aim to improve clarity, relevance, and facility in communication. Analyzing the preposing of other constituents may bring us closer to establishing general underlying principles guiding what has heretofore been seen as ostensibly free word order in Japanese. I l l References Birner, Betty J. and Gregory L. Ward. 1998. Information status and noncanonical word order in English. Amsterdam: John Benjamins Publishing Company. Boscovic, Zdljko. and Daiko Takahashi. 1998. Scrambling and Last Resort. Linguistic Inquiry 29.347-366. Chafe, Wallace. 1976. Givenness, contrastiveness, definiteness, subjects, topics, and point of view. Subject and Topic, ed. by Charles Li, 27-55. New York: Academic Press. . 1980. The development of consciousness in the production of a narrative. The Pear Stories, ed. by Wallace Chafe, 9-50. Norwood, NJ: Ablex. . 1987. Cognitive constraints on information flow. Coherence and Grounding in Discourse, ed. by Russell S. Tomlin, 21-51. Amsterdam: John Benjamins Publishing Company. Chomsky, Noam. 1972. Studies on semantics in generative grammar. The Hague: Mouton. . 1981. Lectures on Government and Binding. Dordrecht: Foris. . 1992. A Minimalist Program for linguistic theory. The Minimalist Program, by Noam Chomsky, 167-217. Cambridge, MA: MIT Press. Clancy, Patricia M. 1982. Written and spoken style in Japanese narratives. Spoken and Written Language, ed. by Debrah Tannen, 55-76. Norwood, NJ: Ablex Publishing Corporation. Clark, Herbert H. and Susan E. Haviland. 1997. Comprehension and the Given-New contract. Discourse Production and Comprehension, ed. by Roy O. Greedle, 1-40. Norwood, NJ: Ablex Publishing Corporation. Collier-Sanuki, Yoko. 1993. Word order and discourse grammar: a contrastive analysis of Japanese and English relative clauses in written narratives. Los Angeles: University of California, Los Angeles, dissertation. 112 Cowan, Ron. 1995. What are discourse principles made of? Word order in discourse, ed. by Pamela Downing and Michael Noonan, 29-49. Amsterdam: John Benjamins Publishing Company. Dryer, Matthew. 1996. Focus, pragmatic presupposition and activated propositions. Journal of Pragmatics 26.475-523. Enchi Fumiko. 1958. Tokyo no tsuchi. Soosaku daihyoo senshuu Vol. 22, ed. by Nihon bungei kyookai, 351-372. Tokyo: Dainihon yubenkai kodansha. Firbas, Jan. 1964. On defining the theme in functional sentence analysis. Trabaux Linguistiques de Prague 1.267-280. Fox, Barbara A. and Sandra A. Thompson. 1990. A discourse explanation of the grammar of relative clauses in English conversation. Language 66.297-316. Frazier, Lyn. and Keith Rayner. 1988. Parameterizing the language processing system: Left- vs. right-branching within and across languages. Explaining Language Universals, ed. by John A. Howkins, 247-279. Oxford: Basil Blackwell Ltd. Givon, Talmy. 1983. Topic continuity in discourse. A quantitative cross-language study. Amsterdam: John Benjamins Publishing Corporation. . 1988. The pragmatics of word-order: predictability, importance and attention. Studies in Syntactic Typology, ed. by Michael Hammond, Edith A. Moravcsik, and Jessica R. Wirth, 243-284. Amsterdam: John Benjamins Publishing Corporation. Gundel, Jeanette K. 1988. Universals of topic-comment structure. Studies in Syntactic Typology, ed. by Michael Hammond, Edith A. Moravcsik, and Jessica R. Wirth, 209-239. Amsterdam: John Benjamins Publishing Corporation. Greenberg, Joseph H. 1966. Universals of Language. 2nd ed. Cambridge, MA: MIT Press. Hale, Ken. 1980. Remarks on Japanese phrase structure: comments on the papers on Japanese syntax. Theoretical issues in Japanese linguistics, ed. by Ann K. Farmer and Yukio Otsu, 185-203. 113 Halliday, Michael A. K. 1967. Notes on transitivity and theme in English, Part 2. Journal of Linguistics 3.199-244. Harada Shinichi. 1977. Nihongo ni'henkei'wa hitsuyoo da. Gengo 6.96-103. Hashimoto Shinkichi. 1948. Kokugohou kenkyuu. Tokyo: Iwanami shoten. Haviland, Susan E. and Herbert H. Clark. 1974. What's new? Acquiring new information as a process in comprehension. Journal of verbal learning and verbal behaviour 13.512-521. Higginbotham, James. 1983. Logical form, binding, and nominals. Linguistic Inquiry 14.395-420. Jackendoff, Ray. 1972. Semantic Interpretation in Generative Grammar. Cambridge, MA: MIT Press. Kato Akihiko, Saji Keizou and Morita Yoshiyuki. 1989. Nihongo gaisetsu. Tokyo: Oufuu. Kuno, Susumu. 1972. Functional sentence perspective: a case study from Japanese and English. Linguistic Inquiry 3.269-320. . 1973. The structure of Japanese Language. Cambridge, MA: MIT Press. . 1978a. Japanese, a characteristic OV language. Syntactic Typology, ed. by Winfred P. Lehmann, 59-138. Austin: University of Texas Press. . 1978b. Danwa no bunpou. Tokyo: Taishuukan shoten. . 1978c. Generative discourse analysis in America. Current trends in textlinguistics, ed. by Wolfgang U. Dressier, 275-294. Berlin, New York: Walter de Gruyter. Lambrecht, Kund. 1994. Information structure and sentence form. Cambridge: Cambridge University Press. Lasnik, Howard, and Tim Stowell. 1991. Weakest crossover. Linguistic Inquiry 22.687'-720. Mahajan, Anoop K. 1990. The A/A' distinction and movement theory. Cambridge, MA: MIT dissertation. Martin, Samuel E. 1975. A reference grammar of Japanese. New Haven: Yale University Press. 114 Masunaga, Kiyoko. 1983. Bridging. Proceedings of the 13th International Congress of Linguists, ed. by Shiro Hattori and Kazuko Inoue, 455-460. Mikami Akira. 1972. Gendaigohou shinsetsu. Tokyo: Kuroshio shuppan. Mithun, Marianne. 1987. Is basic word order universal? Coherence and Grounding in Discourse, ed. by Russell S. Tomlin, 281-328. Amsterdam: John Benjamins Publishing Company. Miyagawa, Shigeru. 1997. Against optional scrambling. Linguistic Inquiry 28.1-25. Miyajima Tatsuo. 1964. Joshi • jodoushi no youhou. Gendai zasshi kyuujyusshu no yougo youji , daisanbunsatsu Bunseki. Kokuritsu kokugo kenkyuujo. Tokyo: Shuuei shuppan. Morita Yoshiyuki. 1989. Imi. Nihongo gaisetsu, ed. by Kato Akihiko, Saji Keizou and Morita Yoshiyuki, 95-106. Tokyo: Oufuu. Murou Saisei. 1958. Kiiroi fune. Soosaku daihyoo senshuu Vol. 22, ed. by Ninon bungei kyookai, 373-392. Tokyo: Dainihon yubenkai kodansha. Okutsu Keiichirou and Tanaka Akio. 1990. Nihongo e no shoutai -bunpou to goi. Kokusai kouryuu kikin. Tokyo: Bonjinsha. Ono, Tsuyoshi. and Ryoko Suzuki. 1992. Word order variability in Japanese conversation: Motivation and grammaticization. Text 12.429-445. Prince, Ellen F. 1981. Toward a taxonomy of Given-New information. Radical pragmatics, ed. by Peter Cole, 223-255. New York: Academic Press. . 1992. The ZPG letter: subjects, definiteness, and information-status. Discourse description, ed. by William C. Mann and Sandra A. Thompson, 295-325. Amsterdam: John Benjamins Publishing Corporation. Reinhart, Tanya. 1979. Syntactic domains for semantic rules. Formal Semantics and Pragmatics for Natural Language, ed. by Franz Guenthner and S. J. Schmidt, 107-130. Dordrecht: D. Reidel Publishing Company. 115 Reinhart, Tanya. 1981. Definite NP anaphora and c-command domains. Linguistic Inquiry 12.605-635. . 1995. Interface strategies, ms. Rochemont, Michael S. 1998. Phonological focus and structural focus. The Limits of Syntax, ed. by Peter W. Culicover and Louise McNally, 337-363. New York: Academic Press. Saeki Tetsuo. 1960. Gendaibun ni okeru gojun no keikou. Iwayuru hogo no baai. Gengoseikatsu vol. 111.56-63. . 1975. Gendai nihongo no gojyun. Tokyo: Kasama shoin. . 1989. Gendai nihongo no tenkai. Tokyo: Izumi shoin. Saito, Mamoru. 1985. Some asymmetries in Japanese and their theoretical consequences. Cambridge, MA: MIT dissertation. . 1992. Long distance scrambling in Japanese. Journal of East Asian Linguistics 1.69-118. Saito, Mamoru. and Naoki Fukui. 1998. Order in phrase Structure and Movement. Linguistic Inquiry 29.439-474. Saito, Mamoru. andHajime Hoji. 1983. Weak crossover and move a in Japanese. Natural Language and Linguistics Theory 1.245-259. Shibatani, Masayoshi. 1990. The languages of Japan. Cambridge: Cambridge University Press. Shimojo, Mitsuaki. 1995. Focus structure and morphosyntax in Japanese: wa and ga, and word order flexibility. Buffalo: State University of New York at Buffalo, dissertation. Takeuchi Kazuko. 1958. Shokujo. Soosaku daihyoo senshuu Vol. 22, ed. by Nihon bungei kyookai, 313-328. Tokyo: Dainihon yubenkai kodansha. Tomlin, Russell S. 1995. Focal attention, voice, and word order. An experimental, cross-linguistic study. Word order in discourse, ed. by Pamela Downing and Michael Noonan, 517-554. Amsterdam: John Benjamins Publishing Company. 116 Ward, Gregory L. 1988. The semantics and pragmatics of preposing. New York: Garland Publishing, Inc. Webelhuth, Gert. 1992. Principles arid Parameters of Syntactic Saturation. New York: Oxford University Press. Whitman, John. 1982. Configurationality parameters, ms. Harvard University, a revised version appears in issues in Japanese linguistics, ed. by Takashi Imai and Mamoru Saito, 351-374. 1987. Dordrecht: Foris. Yagi Yoshinori. 1958. Komatta hibi. Soosaku daihyoo senshuu Vol. 22, ed. by Nihon bungei kyookai, 329-342. Tokyo: Dainihon yubenkai kodansha. 117 APPENDIX A Data Sources Angel, Jeremy. 1992. Happyaku-man-too no aiken jijoo (Man's Best Friend). Translation by Takeuchi Kazuyo. In Igirisu-biiki. Asahi shinbun ronsetsu iin shitsu. 1997. Tenseijingo '97 aki Vol. 110. Tokyo: Hara shobou. Asahi shinbun ronsetsu iin shitsu. 1998. Tenseijingo '97 fuyu Vol. 111. Tokyo: Hara shobou. Edogawa Kouichi. 1992. Takuhaibin no jon renon (True Beatles). In Igirisu-biiki. Ekuni Shigeru. 1992. Ekken! Nigaoe no chaaruzu-denka (An Imperial Audience). In Igirisu-biiki. Endou Shuusaku. 1997. Saigo no hanadokei. Tokyo: Bungeishunju. Hayashi Nozomi. 1992. Shioaji no kiita eikokugun (British Cars). In Igirisu-biiki. Ichikawa Shinichi. 1992. Oyaji-yuzuri no eikoku kiji (Life Props). In Igirisu-biiki. Ofisu masuto Ltd. (Ed.) 1992. Igirisu-biiki. —Aquascutum no bigaku--. Tokyo: Kodansha. Itasaka Gen. 1992. Hajimari wa itsumo Igirisu (Well-spring). In Igirisu-biiki. Kawaguchi Atsuko. 1992. Yasegamanna shinema (Cheers Cinema). In Igirisu-biiki. Kawamura Mikio. 1992. Shaarokian no shubi-han'i (Sharlockian's Sanctuary). In Igirisu-biiki. Miyamoto Michiko. 1994. Itaria tosukaana no yuugana shokutaku. Tokyo: Soshisha. Nishiki Masaaki. 1992. Jon bum no tanken to yamashi no bouken (The John Bull Spirit). In Igirisu-biiki. Nishio Tadahisa. 1992. Otehon wa itsumo misuterii (Gentlemen in Mystery). In Igirisu-biiki. Tawara Machi. 1988. Yotsuba no essei. Tokyo: Kawade shobou shinsha. Ueno Chizuko. 1992. Uwanosora. Doitsu sonohigurashi. Tokyo: Asahi shinbunsha. 118 APPENDIX B List of Definitions A-position: A position to which a theta role such as an agent, patient, theme, etc. can be assigned. A ' (A bar)-position: A position which is not an A-position. Anchored: A newly introduced ("Discourse-new/Hearer-new") NP which is linked to some other discourse entity by means of another NP or 'anchor' properly contained within it. For example, the NP 'a guy I work with' is an Anchored NP containing an anchor T in the following example: A guy I work with says he know your sister. (Prince 1981: 233) Bunsetsu un i t s : Units we first obtain by dividing sentences, and components which immediately compose sentences (Hashimoto 1948). For example, the following sentence consists of eight bunsetsu units. The slashes indicate where the sentence is divided into bunsetsu units. Watashi wa / kinoo / yuujin to / futari de / maruzen e / I TOP yesterday friend with two (people) by Maruzen LOC hon o / kai-ni / ikimashita. book ACC buy-purpose went T went together with my friend to Maruzen to buy a book yesterday.' (Hashimoto 1948: 6) As shown in the above example, adverbs such as kinoo 'yesterday' can themselves form a bunsetsu unit. On the other hand, particles and auxiliary verbs cannot form bunsetsu units by themselves; therefore, in the above example, they form bunsetsu units together with words such as watashi wa T TOP' and yuujin to 'with my friend.' Bunsetsu units in sentences can be identified by inserting elements such as the sentence final particle (SFP) ne or the combination of a copula (CPL) and the SFP desu ne (Mikami 1972) as shown in the following example: (i) inserting the SFP ne: Watashi wa ne, kinoo ne, tomodachi to ne, maruzen e ne, I TOP SFP yesterday SFP friend with SFP Maruzen to SFP shomotsu o ne, . . . books ' ACC SFP (Mikami 1972:27) 119 (ii) inserting a combination of a CPL and SFP desu ne: Watashi wa desu ne, kinoo desu ne, tomodachi to desu ne, I TOP CPL SFP yesterday CPL SFP friend with CPL SFP maruzen e desu ne, shomotsu o desu ne, . . . Maruzen to CPL SFP books ACC CPL SFP (Mikami 1972:27) C-command: A constituent X c-commands a constituent Y if neither X nor Y dominates the other and the first branching node dominating X dominates Y (Reinhart 1979). Clause-internal scrambling: Scrambling in which a constituent other than the subject and the verb appears in a sentence-initial position. Demonstrative: A word which is used to refer to a person or an object in place of the referent's name. A demonstrative locates a referent in relation to a speaker, a referent, and an addressee. Direct object: Someone or something directly involved in an action or process described by a transitive verb. A direct object is marked by the particle o in Japanese; however, not all the NPs marked by the particle o are direct objects. For example, o-marked NPs which indicate a spatial path for motion verbs such as aruku 'walk,' tobu 'fly,' wataru 'cross,' etc. are not direct objects but traversal objects (Martin 1975). Motion verbs such as deru 'leave,' tatsu 'depart,' oriru 'descend from,' etc. also take omarked NPs which are called ablative objects (Martin 1975). Some direct objects are marked by the particle ni, such as taroo ni 'Taro DAT' in the following sentence: Inu ga taroo ni kamitsuita. dog NOM Taro DAT bit 'A dog bit Taro.' (Okutsu and Tanaka 1990: 65) Discourse-new: An NP which is introduced to the discourse for the first time at the time of the utterance. Discourse-old: An NP which has already been evoked in the preceding discourse either textually or situationally. Emphas is : An NP is considered emphasized when it conveys focus by corresponding to the constituent that is wh-questioned in a question (Rochemont 1998). 120 Head: A head determines the range of syntactic functions that a phrase which contains this head can bear. Hearer-new: An NP which is assumed not to be already known to the hearer at the point where it is introduced to the discourse. Hearer-old: An NP which is assumed to be already known to the hearer at the point where it is introduced to the discourse. Idiomatic expressions: Expressions formed with an NP and a verb in which the meaning of the verb differs from its original or literal meaning. Inferrable: A newly introduced NP which can be inferred from Discourse-old entities. For example, the newly introduced NP 'the driver' is inferrable from an already mentioned NP 'a bus' in the following example: I got on a bus yesterday and the driver was drunk. (Prince 1981: 233) IP: A phrase whose head is a verbal inflection (I). Kakari-joshi: Particles that affect the entire predication and figure prominently in the classical language of the Heian period (Shibatani 1990). Kakari-joshi are particles such as wa (topic or contrastive), mo 'also,' koso (emphatic), sae 'even,' demo 'or something,' hoka 'other than,' and shika 'only.' Some theories treat only wa and mo as kakari-joshi. Logica l F o r m (LF) : A syntactic level of representation which encodes logico-semantic properties of a sentence. Long-distance scrambling: Scrambling in which a constituent of an embedded clause appears in the initial position of the clause that contains the embedded clause. Main clause: A clause which can form a sentence independently. Noun Phrase (NP): A phrase whose head is a noun. 121 Scrambl ing: A syntactic term that refers to a phenomenon in which some constituents of a sentence appear in a non-canonical order. Situationally Evoked : A "Discourse-old/Hearer-old" NP which is identified from an entity existing in an extratextual context. For example, the NP 'you' in the following example is a Situationally Evoked NP which refers to a person present when this sentence was uttered: Pardon, would you have change for a quarter? (Prince 1981: 233) Subject: Someone whose role is an agent or an experiencer in a sentence. For example, 'John' is the subject of 'John came.' 'John helped me,' and 'John loves Mary.' Also something which has the same syntactic role and performs the action of the verb, such as 'knowing him' in 'Knowing him helped me.' Subordinate clause: A clause which cannot stand by itself as a sentence but can help to make a larger clause when it is part of, or joined to, a main clause. Textually Evoked: A "Discourse-old/Hearer-old" NP which is identified from an entity already mentioned in the preceding discourse. For example, the NP 'he' is a Textually Evoked NP which refers to the NP 'a guy I work with' mentioned in the preceding discourse: A guy I work with says he know your sister. (Prince 1981: 233) Topic/contrast marker wa: NPs marked with the particle wa comprise either the topic or are contrastive. The following are examples of direct objects marked with wa comprising a topic and wa conveying contrast. (i) Sono hon wa jon ga kaita. that book TOP John NOM wrote 'Speaking of that book, John wrote it.' (ii) Jon wa toriniku wa taberu (ga, butaniku wa tabenai ). John TOP chicken CONT eat but pork CONT doesn't eat 'John eats chicken (but not pork).' When marking a topic, as in (i), wa is used when conveying a general statement about the wa-marked NP and is translated as "as for . . ., speaking of . . ., talking about. . .". When a parallel or contrasting proposition exists overtly or covertly as in (ii), a wa-marked NP conveys a contrast and is translated as "X . . ., but . . ., as for A . . .". 122 Unanchored: A newly introduced ("Discourse-new/Hearer-new") NP which is not linked to some other discourse entity by means of another NP or 'anchor' properly contained within it. For example, the NP 'a bus' in the following example is an Unanchored NP: I got on a bus yesterday and the driver was drunk. (Prince 1981: 233) Unused: A newly introduced ("Discourse-New") NP which is assumed to be already known to the hearer ("Hearer-old"). For example, the NP 'Noam Chomsky' in the following example is an Unused NP: Noam Chomsky went to Penn. (Prince 1981: 233) Variable: X is a variable if (i) X is an A-position (see A-posit ion), and (ii) X is a trace of movement to an A'-position (Saito 1985). Verb Phrase (VP): A phrase whose head (see Head) is a verb. VP-internal scrambling: Scrambling in which the direct object appears to the left of the indirect object within the VP. 123 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0099568/manifest

Comment

Related Items