9A v.0!!! - .v} n I. v‘ I g‘ixn . .. :1 .. {Macy-u. {I If“?! 1 | t i II 0 I... In. [full . .1. flouriltunct. .“i.v .6 . I fil.!..mmol.ao.fiivi a y. .33.;1‘." .1 . fl; m5... 11.2... x it... ‘1. ‘0 ..Jn.f..un,..,...- ....m a I I . “I! i . .. 1;!2‘ .. urflwhovi U9, 1! . A . . . .3': . . , . , . .05).!!{1' ‘40an . . 1|“ t u . .. w , .‘ 758.51,!!! . c . . I , . . . . 3 "I U I! ' '1..- vi“ 3'.” I. {Stir} .liulfi‘..llit ‘- til-.111... It... I all}: '11.! (I. .: turn: 9.1.» , : 4"“? v a Iv 451.4 A .e ‘ ‘ !.l 1.. .. .:.o1! 0' :l . i .. .1.l.!(‘ 1 3.100» 02,1 i.~.'.47l3 .‘flol 1.. 3 {olnf‘llz}. 53.! . . I o . .. l." . I” \v‘! 1!; a .0... , xvl.vl:n 2:9»; 4 .01 .Ai.>§.lliolf A I 1:; .Y x. .o... v.....v.. :3 2... ...f.ol|.v1 1:01;».53? >1 {.0 1.: 1!: ..l' I‘ll‘ . Alt]!!! pl: 4 . t.» -?6v::.-.-.k¢1, I)...‘ . Jilin! -. I. 59.... ‘IOV &.|.‘I£:\uou ”In? ..-I.Aur.l‘ tiff ‘ ll. . .7 . W. 7 ‘ ‘. n .‘i tn l finséqaps “wwrlmwwfifgmmfifiwwfi as, inf. r- -. MIC CHI IGAN STATE IHNHI Illll llfllllHUNHIIUIIHHIHWIW 300916 3340 This is to certify that the thesis entitled THE FUNCTION OF SENTENCE TOPICS IN ORAL AND WRITTEN MANDARIN NARRATIVE presented by Gary Steven Abbott has been accepted towards fulfillment of the requirements for MR degree in Liflngiwc S zit/area Major professor Date /0 jun? icicil 0-7639 MS U is an Affirmative Action/Equal Opportunity Institution LIBRARY {Michigan State UnIversIIy “A — PLACE IN RETURN BOX to remove We checkout from your record. TO AVOID FINES return on or before die due. DATE DUE DATE DUE DATE DUE mes n“ 19% VT— L__JL_]L__ ___|| , i—Tl | MSU Ie An Affirmative ActiorVEquel Opponunlty Institution omens-n1 fl THE THE FUNCTION OF SENTENCE TOPICS IN ORAL AND WRITTEN MANDARIN NARRATIVE by Gary Steven Abbott A THESIS Submitted to Michigan State University in partial fulfillment of the requirements ‘ for the degree of MASTER OF ARTS Department of Linguistics 1992 THE guage UEHCE sente of se then, enCe Manda ABSTRACT THE FUNCTION OF SENTENCE TOPICS IN ORAL AND WRITTEN MANDARIN NARRATIVE by Gary Steven Abbott ‘ Mandarin is often described as a topic prominent lan- guage: a language which places more importance and promi- nence on the sentence topic than the subject of any given sentence. Furthermore, it is wrong to assume that the use of sentence topics in Mandarin is arbitrary. This thesis, then, attempts to determine what discourse parameters influ- ence the function of sentence topics in oral and written Mandarin narratives. To answer this, a quantitative discourse analysis was done on three oral texts and six written texts. Using the model of information flow as presented in Prince 1981, each sentence topic from the texts was identified as reflecting either brand new, evoked, or inferred information. These were then correlated with three other discourse parameters-- animacy, episode boundaries, and topic persistence--in order to determine patterns of sentence topic use. The end re- sults indicate that sentence topics do indeed have charac- teristic functions within a text. Copyright'by Gary Steven Abbott 1992 This thesis is dedicated to my brothers and sisters of Markey Community Baptist Church and Hope United Methodist Church. iv sugge in 'e to Dr But a His w be p0 ACKNOWLEDGEMENTS I would like to formally thank Dr. Abbott for her many 4 suggestions and questions as I wrote this thesis, especially in 'emphasizing’ the need for clarity. I am also grateful to Dr. Preston and Dr. Lin for their time and good counsel. But above all, I give thanks to the Holy Spirit; for without His wisdom, encouragement, and strength none of this would be possible. List n I I FJHIIHII. 123 12222345 . 11111111 p 0 T I I TIC 1 122 222 234.56 N1 12345 S 1234444 2222233333344444444 2222222222222222222 TABLE OF CONTENTS List Of Table’s O O O O O O O O O O O O O I Introduction . . . . . . . . . . . . Purpose . . . . . . . . Defining Topicality . .1 Sentence Topics . . .2 Topic Structures . .3 Discourse Topics . Theoretical Framework The Texts . . . . . Thesis Outline . . II Topicality and Mandarin Syntax . . . IntrOduction O O O O O O O I O O 0 Construction . . . . . . Sentence Topic as Definite . . . ODOIQOON Sentence Topic as Encoding of the Sentence Topic . Lack of a Passive Construction . Absence of Dummy Subjects . . . Double Subject Constructions . . Lack of Subject Verb Agreement . Sentence Topic in Mandarin Syntax Restricting the Spatial Framework 014:.me Restricting the Sentence Topics and Subjects . Absence of Sentence Topics . . NNNNNNNNNNNNNNNNNNN “NH wk h-bnhuwaI-i Sentences . . . . . . . .4 Optional Markers . . . . . . . Summary . . . . . . . . . . . . . 01-h pbhbbebewwwwwwmwmmm NM vi Restricting the Temporal Framework Individual Framewor Other Features of Sentence Topic in Mandarin Characteristics of the Sentence Topic .1 Sentence Topic as a Discourse Related Sentence Topic as Given Information Inferred Information Mandarin as a Topic Prominent Language k Sentence Topic as Pragmatic Knowledge Sentence Topic as Sentence Initial ,Second Sentence Topic in Complex and Compound ix choqmri-s I-‘H 15 15 15 16 17 19 19 22 23 25 27 27 28 28 29 30 32 32 33 33 34 35 36 37 ulll4 1 u q 1“ II C .1]. Lu .1 ne np O - . PS 0 IL AF. T D S D. IL AE T D 111.123 T 123 WW 1123 45 f 112 34 5 66667 12 34. 5 66n6o6~l " C U ' O . 0 - I . D | I 44. 4.4.. 4 4.4.4144 V 55 55 5 55.3205 12222223 . - . .. . . 33333333 33 III COnC III wwwwwwww “NNNNNNH 0000 1 1 .1 2 3 Prince’s Taxonomy of Assumed Familiarity . . . . Introduction . . . . . . . . . . . . . . . . . . Prince’s Taxonomy of Given-New Information . . . New Information . . . . . . . . . . . . . . . .1 Brand New Information . . . . . . . . . . . .2 Unused Information . . . . . . . . . . . . . Evoked Information . . . . . . . . . . . . . . Inferred Information . . . . . . . . . . . . . Applying the Assumed Familiarity Taxonomy to - English . . . . . ... . . . . . . . . . . . Prince’s Taxonomy and Mandarin Sentence Topics . Summary . . . . . . . . . . . . . . . . . . . . IV Topic Choice in Oral Mandarin Narrative . . . . . AA 09-h A“ A .5 .6 .6 .6 .6 .7 ubobérh-P- 'NH .1 .2 3 Introduction . . . . . . . . . . . . . . . . . . Levels of Familiarity and Sentence Topics in Oral Texts . . . . . . . . . . . . . . . . Animacy and Sentence Topics in Oral Texts . . . Episode Boundaries and Sentence Topics in Oral Texts . . . . . . . . . . . . . . . . . J . Topic Persistence and Sentence Topics in Oral Texts . . . . . . . . . . . . . . . . . . . Discussion . . . . . . . . . . . . . . . . . . . Brand New Sentence Topics . . . . . . . . . . Evoked Sentence Topics . . . . . . . . . . . . Inferred Sentence Topics . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . V Topic Choice in Written Mandarin Narrative . . . . VI Introduction . . . . . . . . . . . . . . . . . . Levels of Familiarity and Sentence Topics in Written Texts . . . . . . . . . . . . . . . Animacy and Sentence Topics in Written Texts . . Episode Boundaries and Sentence Topics in Written Texts . . . . . . . . . . . . . . . . . . . Topic Persistence and Sentence Topics in Written Texts . . . . . . . . . . . . . . . . . . . Discussion . . . . . . . . . . . . . . . . . . . .1 Brand New Sentence Topics . . . . . . . . . . .2 Evoked Sentence Topics . . . . . . . . . . . . .3 Inferred Sentence Topics . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . vii 40 4O 4O 41 42 43 44 45 47 48 50 52 52 53 54 57 64 68 68 69 70 71 72 72 73 74 76 80 82 83 84 85 86 88 Append Da Hui Si Mag Dushur1 Heshani Gechani Jixing; Duzou ' Yongsh. Yanli I List of Appendix . . . . Da Hui Lang Text Si Maguang Text . Dushuren Text . . Heshang Text . . Gechangjia Text Jixingzi Text . Duzou Text . . Yongshi Text . Yanli Text . . List of References viii 92 94 126 130 134 145 150 153 156 160 166 Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table 01 ttht-e «16) 10 11 12 13 14 15 16 17 18 LIST OF TABLES Oral Texts--Levels of Familiarity . . . . . . ' Oral Texts--Animacy and Sentence Topics . . . Oral Texts--Animacy and Level of Familiarity Oral Texts--Episode Boundaries and Levels of ~Familiarity . . . . . . . . . . . . . . Oral Texts--Episode Boundaries and Total Occurrences of Levels of Familiarity . . Oral Texts--Episode Boundaries and Animacy. . Oral Texts-—Probabilities and Episode Boundaries . . . . . . . . . . . . . . . Oral Texts--Topic Persistence and Levels of Familiarity . . . . . . . . . . . . . . Oral Texts-~Animacy and Topic Persistence . . Written Texts--Levels of Familiarity . . . . Written Texts--Animacy and Sentence Topics. . Written Texts--Animacy and Level of Familiarity . . . . . . . . . . . . . . Written Texts--Episode Boundaries and Levels of Familiarity . . . . . . . . . . . . . Written Texts--Episode Boundaries and Total Occurrences of Levels of Familiarity . . Written Texts——Episode Boundaries and Animacy Written Texts--Probabilities and Episode Boundaries . . . . . . . . . . . . . . . Written Texts--Topic Persistence and Levels of Familiarity . . . . . . . . . . . . . Written Texts--Animacy and Topic Persistence. ix 53 56 57 61 62 62 63 66 67 73 74 75 76 77 78 79 80 81 1.1 l tics. and ap which import Pragma- narrat; Craftec tic or able to end. N it? is . Coding < the flea However, meaning, Clarify TM is a top: IOpiCS a3 aSSumed INTRODUCTION 1.1 Purpose Topicality is a complex issue in the study of linguis- tics. It has been studied from a variety of perspectives and applied to a variety of genres. One of the conclusions vwhich can be drawn from this diverse field of study is the importance of topicality in the syntax, semantics, and pragmatics of a language. This is especially true in a narrative. A narrative is also a complex entity, and a well crafted one will incorporate many different devices, syntac- tic or otherwise, to ensure that the reader/listener will be able to follow the flow of information from beginning to end. No one device will make or break a text, but topical- ity is certainly of great importance. The effective en- coding of topics can help the reader/listener better follow the flow of information that is presented in the story. However, just as poorly constructed sentences can obscure meaning, poorly encoded topics can confuse rather than clarify a story. This is particularly true for Mandarin Chinese, which is a topic prominent language--as will be illustrated later. This paper, then, will examine the realization of Mandarin topics as they relate to an informational taxonomy of assumed familarity. Additional factors such as animacy, 1 episor I indica entire the di topic demons certair ratives will al Th Section of 'tOp Confusi theoret this pa] commentS analysiE remainde 1'2 Def Top UnfortUm lng What 2 episode boundaries, and topic persistence (a numerical indication of the relative importance of a topic to the entire discourse) will be examined to determine what are the discourse rules which are involved in the use of one topic over another. It is the purpose of this thesis to demonstrate that the encoding of topics is conditioned by certain pragmatic factors. Since oral versus written nar- ratives reflect different mediums, the encoding of topics will also follow different, albeit minor, discourse rules. The remainder of this chapter will be divided into four sections. The first will differentiate among various uses of 'topic’ and to define those uses so as to minimize any confusion in how ‘topic’ is used in this thesis. Next, the theoretical framework used in the analysis of the data for this paper will be presented. In the third section some comments will be made about the texts that are used in this analysis.‘ The final section will present an outline of the remainder of this thesis. 1.2 Defining Topicality Topicality is a popular area of study in linguistics. Unfortunately, ‘topic’ can be a confusing term, and defin- ing what 'topic’ means can be a challenge.1 Hakulinen (1989: 62) suggests several possibilities, but ends by rejecting them all. Other linguists, like Levinson (1983:x), feel that there is so much confusion about what is meant by 'topic’ that any discussion about topicality is pointless. 3 Often this confusion results from the use of the same term to refer to related but different aspects of topicality from one discussion to the next. By differentiating between these various domains and applying different terms to refer to each, much of this confusion can be minimized. There- fore, 'topic’ will be used to cover three distinct but closely related areas: sentence topic, topic structure, and discourse topic. 1.2.1 Sentence Topics. The first, sentence topic, is perhaps the most well known of the three, and often is de- fined as what the sentence is about (Davison 19842804). From this definition it is safe to say that most any sen- tence in any language will have at least one primary sen- tence topic. To suggest otherwise would be to claim, para- doxically, that a sentence is not about anything, which is difficult to envision. As a result, the sentence topic of the_sentence I like beans can be one of three different topics--the speaker who likes beans, the speaker’s attitude towards beans, or beans which the speaker 1ikes--depending upon which aspect of the sentence the speaker wishes the sentence to be about. Obviously, context is of primary importance in determining what the true sentence topic of any given sentence is. Here, it is useful to make a distinction between the concept of 'sentence topic’ as opposed to the concept of 'focus.’ Both are used to describe what a sentence is about, but they do not refer to the same type of infor thesi ticipa the 0H The 01 by the Object B) and Where the SI all t] 4 information. A sentence topic, for the purposes of this thesis, reflects old or given information of which the par- ticipants in the discourse are already aware. Focus, on the other hand, reflects new or reactivated information. A. What do you like? B. I like beans. The old information in B’s response is B, which is encoded by the pronoun 'I.’ The new information is beans, as the object of what he likes. Therefore, the sentence topic is B, and the focus is beans. ' Focus is a useful notion in the study of conversation, where new information gains prominence in reflecting what the speaker is talking about. In the following dialogue all the focused information is in italics. A. Speaking .f beans, I like them. B. I don’t like them at all. I much prefer avocados. A. Isn’t guacamole made from avocados? B. Yeah, and my friend Dorothy makes the best guaca- mole in the world. Here, it is clear that the focused information is not the same from one speaker to the next. Quite often, what was focused, new information, in one utterance, becomes old, nonfocused information in the following utterance. In the rapid give and take of conversation, it is very possible for the focus to change from one speaker to the next. As a result, focused nouns very often are low in topicality-~that 5 is, they are not persistent in the text. They may occur in several clauses, but then are dropped altogether. Sentence topic, on the other hand, is very useful in the study of a narrative. Unlike conversation, a sentence topic in a narrative is more likely to be repeated (often with zero or pronominal reference) over several clauses. This is useful as it provides a familiar framework for the participants to process new information. In the following mini-narrative, all of the sentence topics are in italics. I like beans. Beans are a versatile food. They are high in protein, and ¢ are inexpensive. Plus, they can be prepared in a variety of ways. Here, beans, as old information, is used repeatedly as a framework to present a variety of new information. It is this sense of a sentence topic reflecting old information which will be used in this thesis. Therefore, in order to more clearly distinguish sen- tence topic from focus, the definition of a sentence topic, as given by Davison as what the sentence is about needs to be revised. A sentence topic, then, is given or available information which "sets a spatial, temporal, or individual framework within which the main predication holds" (Chafe 1976:56). Note that a sentence topic will still be one of the things that the sentence is about, but it now reflects old-given information not new. Later, when Mandarin sen- tence topics are discussed, this definition will be illus- trated in greater detail. tic a topic topic of do; markeo But a initia struct the Se Struct howeve T tween tion. makes ture i a Drag 6 1.2.2 Topic Structures. Topic structure is the syntac-> tic and sometimes morphological encoding of the sentence topic by some word or phrase which refers to the sentence topic. Like sentence topic, most languages have some means of doing this. Quite often a sentence topic is more clearly marked by placing it in a sentence initial construction. But a topic structure does not always have to be sentence F- initial, and the encoding of the sentence topic into a topic structure will vary from language to language. In Mandarin, the sentence topic does occur in a sentence initial con- struction, as will be shown later in chapter two. English, however, has no such restriction. This difference in the encoding of a sentence topic be- tween Mandarin and English can present problems in transla- tion. The sentence initial topic structure of Mandarin. makes it tempting to employ a sentence initial topic struc- ture in English. But this can result in an awkward or even a pragmatically incorrect translation. 1. Da Hui Lang text, line 42 Lao nainai, toufu hen chang . . . old grandmother hair very long The old grandmother had hair that was very long A more literal translation of the above sentence--The old grandmother, her hair was very long--preserves the topic structure as used in Mandarin, but it is not a pragmatically accurate translation of the Mandarin. This is because this type of topic structure in English carries not only an eleme also tivat handa persi. Thompe langu topic nent l tOpic 1 Course the te) about), is abou Cally, t0Pic, known a known 8 Ta Da HUi COUPse in this OCCUFI‘i 7 element of contrast not present in the Mandarin sentence, it also is a construction more commonly used for new (or reac- tivated), focused, non-persistent information. But in the Mandarin, this same topic structure is used for old, highly persistent information. This has led some linguists (Li and Thompson 1976) to make a distinction between topic prominent languages (those that consistently use sentence initial topic structures in this nonfocused use) and subject promi- nent languages (those that do not). 1.2.3 Discourse Topics. Lastly, discourse topic is a topic with a wider scope than that of the sentence. Dis- course topic can refer to a variety of different levels in the text-éfrom the entire text (what the whole narrative is about), to episodic units within the text (what the episode is about), to smaller units within the episode. Specifi- cally, when several nearby sentences have the same sentence topic, the overriding topic for a group of such sentences is known as the discourse topic, and the group of sentences, is known as the topical span. Take, for example, the three lines given below from the Da Hui Lang text. The sentence topic, and also the dis- course topic, is indicated in italics (the two little girls in this text are old information--the last previous mention occurring in line 102). The ini ”"989 t identif identif: °f the .; "eakEr 1 noun, (t IhEre is This is first me: I mien, a null r“ 8 2. Da Hui Lang text, line 104 Zhe liang ge xiao nuhai gen lao nainai this two CLSF small girl with old grandmother Yiqi chang da. together grow big These two little girls with the old grandmother grew up. 3. Da Hui Lang text, line 105 Tamen guo de feichang xinfu. 3pl live MNR very happily They lived very happily. 4. Da Hui Lang text, line 106 Tamen yizhi zai shan li zhu, ¢ zhu de 3pl all the time at hill in live live MNR feichang hao. very good All the time they lived in the hills, and lived very well. The initial mention of the discourse topic in line 104-- these two little girls with the old grandmother-~is a strong identification, using full noun phrases which specifically identify each of the characters involved. The next mention of the discourse topic in line 105, and also in 106, is a weaker identification, using the third person plural pro- noun, ‘tamen’ they. And also in the second clause of 106 there is the weakest possible identification--zero anaphor. This is a very typical pattern for a topical span. The first mention of the discourse topic is a strong identifi- cation, while successive mentions of the discourse topic use progressively weaker identifications-—even to the point of a null reference, or zero anaphor, if allowed. 1.3 and w does Bindi synta gramme (1974l (1988, approa narrat 1.3 Theoretical Framework This thesis will involve a discourse analysis of oral and written Mandarin narratives. Though discourse analysis does not exclude more formal theories like Government and Binding, the focus of this paper will not be to determine syntactic constraints on the co-occurrence of certain grammatical items as they relate to topicality as Teng (1974), Huang (1982), Xu and Langendoen (1985), and Shen (1988) do. Rather, this thesis will apply a functional approach in analyzing the discourse structure of the narratives. The central aim of a functional theory of syntax is to determine "the relation between linguistic form and_lin- guistic function" (Croft 1990:17). Functionalism, then, is founded on the premise that languages encode semantic and pragmatic information into the syntactic structure of an utterance. The key to understanding a language and how it works is to examine how the structure of the language func- tions to help the discourse participants recover the perti- nent semantic and pragmatic intent of the speaker. In English, for example, the most common sentence pattern-~the unmarked sentence pattern--used to convey basic straight- forward information is the SVO pattern. Thus, I like beans is the simple assertion of the speaker’s fondness for beans. Since subject position in English is typically also the sentence topic (Givon 1984:140-41), the utterance will be understood to be about the speaker. this emplo such const tion under. the Y means very i Struct Entity by ICC} these . text 81 Ple, i. rePI‘eSe indefir the Sut that re Therefo‘ 10 However, if the speaker wishes to make the object of this utterance--beans-—the focus, he or she will need to employ one of the several focus structure constructions, such as Beans I like. This less typical, more marked construction works, then, to encode the pragmatic informa- tion that it is the object of the utterance which is to be understood as the focus of the utterance. That is to say the Y Movement construction of Beans I like functions as a means to mark the object as the focus. From this perspective, it is clear that context is a very important factor in influencing the use of one con- struction over another. But context is a rather difficult entity to pin down and analyze in and of itself. It is only by focusing on certain specific parameters and analyzing how these interact that any insight can be gained into how con- text affects the use of one form over another. For exam- ple, it has been noted that in English, noun phrases which represent new information in the discourse are usually indefinite and will occur in the object position, whereas the subject position tends to have definite noun phrases that represent old information. 5. What did you do today? a. I ate a banana and an orange. But the banana was spoiled. b. ?I ate the banana and the orange. ?But a banana was spoiled. Therefore, the parameters of +/- definite and +/- new can be us in ob. claim course A If 5—? H 54‘ l . C: m U) :1 m r—fi What r will a M m 11 texts , by thr. range collegfi Each H which SQPibe word t 11 be used to predict the likelihood of a noun phrase occurring in object or subject position. From this it is possible to claim that the function of the object is to introduce new information into the discourse encoded as indefinite phra- ses, while the function of the subject is to maintain old information in the discourse encoded as definite phrases. Myhill (1992zch.1, 4) sums up this approach to dis- course analysis in the following way: The basic purpose of typological discourse analysis, then, is to provide a framework for the description of “factors affecting alternations between different forms and constructions in languages in general. The para- 'meters used in this description are quantitative, and statements are made along the lines of 'when property X (e.g. human agent) is present, construction A is used a certain percent of the time, while when property Y (e.g. non-human agent) is present, construction A is used a certain percent of the time.’ What remains to be seen is what the parameters are which will affect the encoding of sentence topics in Mandarin. 1.4 The Texts The analysis in this paper is based upon three oral texts and six written texts. The oral texts were related by three different native Mandarin speakers. Though they range in age from mid-thirties to seventies, they all are college educated-and all came from the area of Shanghai. Each was asked to orally relate three stories in Mandarin, which were recorded. Another native Mandarin speaker tran- scribed the recorded stories into pinyin and gave a word for word translation from the Mandarin into English. From these nine : analys nese Handar T strate the te are al OCCUI‘ 12 nine stories, three, one from each narrator, were chosen for. analysis. Five of the written texts are from a book of Chi- nese folktales (Qin 1983). One was given to me by a native Mandarin speaker, who is also from the Shanghai area. Throughout the paper, these texts will be used to illu- strate important points as the need arises. Each line from the texts has a word for word gloss. Numerous abbreviations are also used to identify the syntactic categories which Below is a list of those abbreviations. occur in each line. ls First Person Singular 1pl First Person Plural Zs Second Person Singular 2pl Second Person Plural 38 Third Person Singular 3pl Third Person Plural ADV Adverbializer CLSF Classifier CRS Currently Relevant Status--Perfect Aspect DO Direct Object Marker DUR Durative Aspect EXP Experiential Aspect GEN Genitive INTR Interrogative MNR Manner PAS Passive Marker PRF Perfective Aspect PRTL Sentential Particle RLV Relativizer RST Resultative 0 Additionally, for ease of reference, each time a specific line from a text is cited the name of the text it came from and the line number within the text will be given. 6. Si Maguang text, line 5 Zhe xuduo xiaohai dou jingya 1e. this many child all surprise PRF This surprised all of the children. Each Final Since it w01 Manda”r USed 13 Each text can also be found in its entirety in the Appendix. Finally, tone is not indicated on any of the Mandarin words. Since tone is not a relevant issue in this paper, to include it would needlessly complicate the examples. 1.5 Thesis Outline This thesis will include six chapters. The first has primarily dealt with defining the various uses of ’topic’ in this thesis, and the theoretical approach that is taken in analyzing the data. The second chapter concerns topicality and Mandarin syntax. In order to gain a better insight into what a sentence topic is, several prototypical characteristics are given and explained. Following this is a discussion of why Mandarin is considered a topic prominent language. Finally, the syntactic features of how Mandarin sentence topics are encoded is given. The third chapter presents Ellen Prince’s (1981) taxo- nomy of assumed familiarity. As such, much of the chapter is devoted to explaining the differences between her three levels of textual information: brand new, evoked, and in- ferred. The last part of the chapter concerns how Prince’s taxonomy can be applied to Mandarin sentence topics. The fourth chapter presents the analysis of the oral Mandarin texts. Each will be broken down into the percent- age of brand new, evoked, and inferred sentence topics used. Additionally, other parameters will be examined to determine how they may or may not influence the use of brand new, inclu But in analy; The 82 used i chapte encodi on why Pics 1. 14 new, evoked, or inferred sentence topics. These parameters include animacy, episode boundaries, and topic persistence. The fifth chapter is exactly the same as the fourth. But in this chapter, the written Mandarin narratives will be analyzed to determine what affects sentence topic encoding. The same parameters will be used in this analysis that are used in chapter four. lAdditionally, at the end of this chapter there will be a discussion on how sentence topic encoding differs from oral to written narrative. The sixth and final chapter will offer some conclusions on why there are differences in the encoding of sentence to- pics in oral and written narrative. TOPICALITY AND MANDARIN SYNTAX 2.1 Introduction Although topicality has been divided into three domains and a definition for each of these has been given, it is still necessary to have a better understanding of both sen- tence topic and Mandarin syntax before any analysis of topic encoding is attempted. The following three sections will provide this information. The first will discuss specific characteristics of the sentence topic. Following that will be an explanation of why Mandarin is considered a topic prominent language. Next will be a section on how to ident- ify sentence topics in Mandarin. A final section will summarize the information presented in this chapter. 2.2 Characteristics of the Sentence Topic Earlier a definition was given for a sentence topic as the identifying framework for the sentence. Though this definition is adequate in describing what a sentence topic is, definitions in and of themselves often lack a list of pertinent characteristics which are actually more useful in identifying a good example of the defined object. The definition for 'cat’ is "a common domestic mammal long kept lmy man as a pet or for catching rats and mice" (Woolf 1974:121). Using this definition alone, it would be rather 15 purrs ident typic examp istic then 16 difficult to distinguish a cat from other domesticated mammals. But if told that a cat is an animal that has pointed ears, long whiskers, a rough tongue, and which purrs, mews, and hisses then a cat becomes much easier to identify. Of course, the danger with any list of proto- typical characteristics of an object is that not every example of the object will have all of the listed character- istics. However, as long as this danger is kept in mind, then these characteristics can be very helpful. 2.2.1 Sentence Topic as a Discourse Related Construc- tion. Perhaps the most important characteristic to keep in mind about sentence topics is that they are discourse re- lated constructions. That is to say, the sentence topic functions in such a way as to allow the "efficient matching of a sentence to context" (Davison 1984:841). The sentence topic creates a link between the sentence it is a part of and the context in which it occurs. As such, it allows the discourse participants to focus on the relevant contextual setting. This is why it makes no sense for a person to walk up to a stranger in a department store where there are no beans in sight, and say I like beans. There can be no con- text that the person spoken to is aware of which he or she can relate the sentence topic to. Rather than encouraging a dialogue, this type of behavior is liable to get odd looks and hurried exits. Even if a discourse context has been created, the sen- tence topic still must be relevant to the ongoing discourse. 1:. It1 dle pet: cont linl lowi spea rela* discc must ticip this Presu tean infer] then ; tenCe CiPant begs f discou: sci°Usz “it, 17 It would still make no sense to say I like beans in the mid- dle of a discussion about the merits of cats as household pets. Provided that beans have not been mentioned in the context, the sentence topic cannot serve as the discourse link that it ought to functions as. However, in the-fol- lowing dialogue beans have been mentioned, which allows speaker B to to use it as a link in his utterance. 1. A: My cat will eat anything, even beans. B: Beans are very tasty. I especially like baked beans. 2.2.2 Sentence Topic as Pragmatic Knowledge. Closely related to the characteristic of the sentence topic as a discourse link is the requirement that the sentence topic must in some way "belong to the pragmatic knowledge of par- ticipants" (Hannay 1985:51) in the discourse. Gundel makes this same assertion when she says that sentence topics are presupposed information (Gundel 1977:30). Unless the sen- tence topic has been previously introduced or represents information that is always accessible to the participants, then it will fail as a discourse link. 1 Earlier it was noted that beans could not be the sen- tence topic in a discussion about cats. Even if the parti- cipants in the discourse were aware of the speaker’s fond- ness for beans, beans have not yet been introduced into the discourse. As such, beans are not part of the current con- Sciousness or pragmatic knowledge of the participants, nor is it presupposed information. This means that typically It w dle pets cont link lowii speal relat disco must ticip this Presu tean infor; then tenCe cipanw neSs 1 diam Scious is it 17 It would still make no sense to say I like beans in the mid- dle of a discussion about the merits of cats as household pets. Provided that beans have not been mentioned in the context, the sentence topic cannot serve as the discourse link that it ought to functions as. However, in the fol- lowing dialogue beans have been mentioned, which allows speaker B to to use it as a link in his utterance. 1. A: My cat will eat anything, even beans. B: Beans are very tasty. I especially like baked beans. 2.2.2 Sentence Topic as Pragmatic Knowledge. Closely related to the characteristic of the sentence topic as a discourse link is the requirement that the sentence topic must in some way "belong to the pragmatic knowledge of par- ticipants" (Hannay 1985:51) in the discourse. Gundel makes this same assertion when she says that sentence topics are presupposed information (Gundel 1977:30). Unless the sen- tence topic has been previously introduced or represents information that is always accessible to the participants, then it will fail as a discourse link. Earlier it was noted that beans could not be the sen- tence topic in a discussion about cats. Even if the parti- cipants in the discourse were aware of the speaker’s fond- ness for beans, beans have not yet been introduced into the discourse. As such, beans are not part of the current con- sciousness or pragmatic knowledge of the participants, nor is it presupposed information. This means that typically a se not the into late twee Beca alloi topic the j actUE infor note Canno .18 a sentence topic must first be introduced into the context, not as a sentence topic, but as part of the comment. Once the information has been inserted into the discourse, and into the pragmatic knowledge of all participants, it can later be used as a sentence topic. The above dialogue be- tween speaker A and B is a good illustration of this point. Because speaker A introduces beans into the discourse, this allows speaker B to use this information as a sentence topic. Of course, in doing so, speaker B may change even the larger discourse topic from cats to beans. This can also be illustrated from an example from an actual text. In the Dushuren text, line 3 contains the new information that a rope was used to tie some books together; note that the rope in this line is new information and so cannot be the sentence topic of this line. 2. Dushuren text, line 3 Ta dai 1e hen duo shu, yong shengzi kun hao 33 take PRF very many book use rope tie well 1e you shutong bei zai jian shang, PRF by servant carry at shoulder on yao jin cheng. want enter city He had many books, which were tied together with a rope, which his servant carried on his shoulder. In line 11, the rope is used as a sentence topic. 3. Dushuren text, line 11 Shengzi duan 1e. rope break PRF The rope broke. The pre informa be used 2. sentenc the cori positio want th. overall the disc tOIlowS Previou in 8 set initial is Perti 2.: Sentenc. prEVIOu; t0Pic S establi, discOUr English of info fUrther 19 The previous mention of Shengzi in line 3 establishes this information in the discourse. This allows Shengzi to later be used as the sentence topic in line 11. 2.2.3 Sentence Topic as Sentence Initial. Since the sentence topic functions as a link between the sentence and the context, it almost always occurs in a sentence initial position (Moutaouakil 1985:83). It makes logical sense to want that part of the sentence which serves as a link to the overall context to occur early in the sentence. This allows the discourse participants to know how the information which follows the sentence topic is related to what has happened previously in the discourse. To place the sentence topic in a sentence final position invites the possibility of initial confusion as to how the information in the sentence is pertinent to the wider context. 2.2.4 Sentence Topic as Definite. The fact that a sentence topic almost always is information that has been previously introduced will naturally require a definite topic structure. As Chafe claims one of the best ways to establish definiteness is "through prior mention in the discourse" (Chafe 1976:40). This is a very common method in English as well as in Mandarin. The first mention of a unit of information is often as an indefinite noun phrase, but further mention will be with a definite noun phrase. 4. A friend came to dinner last night and brought a pot of beans. I was pleased to see that the pot of beans contained baked beans and not lima beans. In sent duced u informa mention Process Maguang exiSten diSCOur “(“1“ pi. line, t 01d’ 8.! gang :1 a Sent. Dartic have b tie“ t is Chg 20 5. Si Maguang text, line 3 Zuiwan de {shihou, you yi ge xiaohai hurang playing RLV time have one CLSF child suddenly duo jin shui gang li . . . fall into water container in While they were playing, a child suddenly fell into a large water container . . . 6. Si Maguang text, line 4 Zhe ge shui gang zhuang mang le shui. this CLSF water container contain full PRF water This water container was full of water. In sentence 4, the information of a pot of beans is intro- duced using an indefinite noun phrase. Once introduced, the information is now established into the discourse and later mention can be done using a definite noun phrase. This same process works in Mandarin as well. In line three of the Si Maguang text, a child falls into a water container. The ‘existence of the water container is new information to the discourse and is entered into the text through an indefinite noun phrase shui gang 'a water container’. In the next line, the water container is no longer new information but old, and is encoded by a definite noun phrase zhe ge shui gang 'this water container’. This same process is true for sentence topics. Since a sentence topic must be presupposed information for the participants, the phrase encoding the sentence topic will have been previously entered into the text as new informa- tion through an indefinite noun phrase. When the entity is chosen for.a sentence topic, the topic structure used for the can be as show It amples full de the sen informa any num anaphor Pronoun but is 1 Re moSt rec then c01 1992; c} the last greater full not The Smal 21 for the sentence topic will, therefore, be definite. This can be seen in lines three and four of the Si Maguang text, as shown above. It should not be construed that based on the above ex- amples every nominal sentence topic be will encoded as a full definite noun phrase. As noted earlier in chapter one, the sentence topic is an entity which represents old-given information. As such, a sentence topic can be encoded by any number of definite phrases including pronouns and zero anaphor. Very typically, the use of a full noun phrase, a pronoun, or zero anaphor is not simply a matter of choice but is more often related to referential distance. Referential distance is determined by identifying "the most recent previous mention of the referent of the NP and then counting how many clauses back it occurred" (Myhill 1992: ch. 2, pg. 14). The fewer the number of clauses to the last mention, the lower the referential distance. The greater the referential distance, the more likely that a full noun phrase will be used for a nominal sentence topic. The smaller the referential distance, the more likely a pronoun or zero anaphor will be used. Therefore, a sen- tence topic which is repeated for several sentences will not very likely be encoded each time with a full noun phraSe. The first mention very likely will be a full noun phrase, but each successive mention will be with a pronoun or zero anaphor. This was illustrated in examples 2-4 in chapter one with the Da Hui Lang text (lines 104-106). Ho to pre Both Fl importa phrase re-enst occurre anaphor Th nite, t, certain then, w first m; tance is mention. ePiSode 2.2 related topiCs l 22 However, referential distance is not always a safe way to predict the occurrence of a pronoun or zero anaphor. Both Flashner (1987:143) and Tomlin (1987:457) recognize the importance of episode boundaries in the type of definite phrase used. A definite full noun phrase will be used to re-enstate the sentence topic after an episode boundary has occurred. Within the actual episode, a pronoun or zero anaphor will be used after the initial mention. Therefore, although a sentence topic is always defi- nite, the definite phrase it is encoded in is sensitive to certain discourse parameters. A nominal sentence topic, then, will be encoded as a full noun phrase, if it is the first mention in a new episode, or if the referential dis- tance is sufficiently high enough to warrant a more explicit mention. A pronoun or zero anaphor will be used within an episode and if the referential distance is relatively low. 2.2.5 Sentence Topic as Given Information. Closely .related to use of definite noun phrases to encode sentence topics is the fact that sentence topics commonly represent given information. This, of course, goes back to prior men- tion. Since sentence topics are information that has al- ready been mentioned, it follows that they no longer convey new information--but rather given information. Taken all together, this means that informational entities in the text with characteristics of "previous mention, giveness, and definiteness" are likely candidates for sentence topics (Bolkestein 1985z3). 2 also pc if it If thi common (Hannay sentenc earlier accepta The inde T0 have dual egi °f Part these 9% sentEnCe Thi the firs fOur dau. 23 2.2.6 Sentence Topic as Inferred Information. It is also possible to use information as a sentence topic even if it has not been previously introduced into the discourse. If this information can be inferred ' 'via logical—-or, more commonly, plausible--reasoning from entities already evoked" (Hannay 1985:52-3), then this information can be used as a sentence topic even if no explicit mention has occurred earlier. For example, the following mini-discourse is acceptable: 7. The other day, I got some eggs. One of those eggs was rotten. The indefinite phrase 'eggs’ is used in the first sentence. To have some eggs implies that there are a number of indivi- dual eggs which comprise the bunch. Since this relationship of part to whole can be inferred, the mention of one of these eggs is not new information, and so one egg can be the sentence topic of the next sentence. This very same process also occurs in Mandarin. In the first line of the Da Hui Lang text, the narrator gives the information that there is a family, and that family has four daughters. 8. Da Hui Lang text, line 1 You yi jiaren, ta jia li you si ge Have one family 33 family in have four CLSF nuer. daughter There was a family which has four daughters. Later, the ser specifi 24 Later, in line 10, the two smallest daughters are used for the sentence topic even though no explicit mention of these specific characters has occurred. 9. Da Hui Lang text, line 10 Liang ge xiao nuer bu dongshi. two CLSF small daughter no understand The two smallest daughters did not understand. Since the fact that there are four daughters has already been mentioned, it is logical to assume that these daughters range in age from youngest, younger, older, oldest. Only in an unusual case where there are twins, triplets, or even quadruplets would this type of inference not be allowed. But this scenario is so unusual the narrator would be obliged to mention it. Therefore, the information that there are two younger daughters can be inferred from the ex- plicit previous mention of the broad information of four daughters-—again, a part to whole relationship. This, then, permits the two youngest daughters to be the sentence topic. From the above discussion, the typical characteristics of a sentence topic can be listed as follows: A. Sentence topics have a discourse function of linking the sentence to the context. B. Sentence topics will occur sentence initially. C. Sentence topics will belong to the pragmatic knowledge of the participants. D. Sentence topics will be encoded by definite phrases in the topic structure. E. Sentence topics will either be information previously mentioned in the discourse. F. Or, sentence topics will be inferrable from other information in the discourse. 25 Again, these are prototypical characteristics, and every sentence topic may not have all of them. However, the vast majority of sentence topics in both English and Mandarin (as well as other languages) will. 2.3 Mandarin as'a Topic Prominent Language Throughout this discussion of typical sentence topic characteristics, several examples from Mandarin have been used as illustrations side by side for the most part with examples from English. But this should not be taken to mean that English and Mandarin use sentence topics in the same way. In fact, there is such a difference in their re- spective uses that English has been termed a subject prom- inent language and Mandarin a topic prominent one (Li and Thompson 1976:460). This is most readily seen in the translation of a Mandarin sentence. 10. Yanli text, line 7 "Zhe bu shi xiezhe 'guangming zhengzhi’ si this no is write guangming zhengzhi four ge da 21 ma!" CLSF big character INTR "Aren’t the characters 'guangming zhengzi’ writ- ten in big characters!" 11. Yanli text, line 8 "'Guangming zhengzi’ si ge zi you guangming zhengzi four CLSF characters have dou name da, shei kanbujian? all like that big who cannot see "Who cannot see those four big characters?" The set zhengz. repress line 7, sentend about t Er The abc guangmr HGuana cannot ted tra see the not Ved tiVeneS too v10 to move does no those f English Mandari latiOn i Bu JUSt be. topics 1 handari that L1 26 The sentence topic of line 8 of the Yanli text is guangming zhengzi. This can be determined because guangming zhengzi represents old information--it had a previous mention in line 7, and it sets a framework for the remainder of the sentence--the rest of the sentence provides new information about these four characters. English has a difficult time smoothly translating this. The above translation does not even include the phrase guangming zhengzi. A more literal translation would be ’ the four characters are all big, who "Guangming zhengzi, cannot see that?’ But this is rather awkward. A less stil- ted translation is 'As for 'guangming zhengzi’, who cannot see that those four characters are big?’ But this too is not very satisfactory and also has the element of contras- tiveness. Other focus constructions are possible, but these too violate the notion of sentence topic as used here. Even to move the sentence topic out of sentence initial position does not help much: 'Who cannot see that guangming zhengzi, those four characters, are all big?’ The difficulty is that English does not have an easy way to carry over both the Mandarin sentence topic and subject into a smooth trans- lation without suggesting focus. But to simply claim that Mandarin is topic prominent just because it is able to use sentence initial sentence topics more freely than English is not conclusive. However, Mandarin syntax does adhere to many of the characteristics that Li and Thompson (1976:466-471) and Fuller and Gundell (1987: lar, He tures: tence i teatUre with Ma 2. lan8Uag tEXtS U! Ellglish1 paSSiVe‘ 27 (1987:4-8) list for topic prominent languages. In particu- lar, Mandarin holds to the following topic prominent fea- tures: a specific encoding of the sentence topic, lack of a passive construction, absence of dummy subjects, double subject constructions, and lack of subject-verb agreement. 2.3.1 Encoding of the Sentence Topic. In Mandarin, the sentence topic is always encoded as a sentence initial phrase--apart from conjunctions, vocatives, and other dis- course support markers. This can be seen in the above exam- ple from the Yanli text. English can also encode the sen- tence topic, and does so in even more ways than just a sen- tence initial topic structure. However, this is the only feature of topic prominent languages that English shares with Mandarin. 2.3.2 Lack of a Passive Construction. Topic prominent languages typically lack a passive construction, or if they have one it is an infrequent construction. In the nine texts used in this thesis, consisting of over 200 lines, only once is a passive construction used (0.5%). The sole occurrence is found in the Gechangjia text, line 8. 11. Gechangjia text, line 8 Xue Tan bei laoshi de gesheng jingdai le. Xue Tan PAS teacher GEN voice stun PRF Xue Tan was stunned by the teacher’s voice. English, on the other hand, makes more frequent use of the passive. Specifically, in works of fiction, like the texts used in this thesis, English passives occurred 9% of the time (C is not increas 2. inent I Since t they do does. left on Portanc Therefo non-ref Where t SUbiect one and fOllowi lanSUagq tioh. i which is Bible tc thr°ush Th°Ush E 28 time (Givon 1990:573). Certainly, in and of itself, this is not significant. But it represents a rather dramatic increase of usage compared to Mandarin narratives. 2.3.3 Absence of Dummy Subjects. Again, topic prom- inent languages do not have dummy or subject constructions. Since topic prominent languages promote the sentence topic, they do not pay as much attention to the subject as English does. ‘If a sentence does not need a subject, it is simply left out. English, on the other hand, places much more im- portance on the subject and little on the sentence topic. Therefore, it will require a subject even if that subject is non-referential. This is seen in the following example where the Mandarin sentence does not have a non-referential subject, but where the English translation at least allows one and probably prefers one. There is no way to say the following Mandarin sentence by using a dummy subject. 11. Da Hui Lang text, line 9 Shan shang you hen duo xian hua. mountain side on have very many fresh flower On the mountain side (there) are many fresh , flowers. 2.3.4 Double Subject Constructions. Topic prominent languages use what is termed a "double subject" construc- tion. Here, there is a sentence initial sentence topic which is followed by the subject. However, it is not pos- sible to derive every occurrence of these constructions through some movement rule from a more basic sentence type. Though every topic prominent language uses these double subject PA 2. topic p between to the in tOpi SUbiect 29 subject constructions, no subject prominent language does. 12. Da Hui Lang text, line 42 Lao nainai, toufu hen chang . . . old grandmother hair very long The old grandmother had hair that was very long. *The old grandmother, hair is very long. 2.3.5 Lack of Subject-Verb Agreement. Finally, in topic prominent languages, there is rarely any agreement between the subject and the verb. Again, this goes back to the importance of the sentence topic over the subject in topic prominent languages. Mandarin is a case in point. Subjects do not agree with the verbs in number or person. 13. Da Hui Lang text, line 48 Zhe shi da hui lang zhu de difang. this is big grey wolf live RLV place This is the place that a big grey wolf lives. 14. Da Hui Lang text, line 106 Tamen yizhi zai shan li zhu . . . 3pl all the time at hill in live All the time, they live in the hills . . . In the above two sentences, the verb zhu 'to live’ occurs first with a third person singular noun da hui lang 'big grey wolf,’ and in the second sentence with a third person plural noun tamen 'they.’ But unlike the English, which adds an agreement marker to the verb, Mandarin does not. Taken all together, Mandarin is a language that encodes the sentence topic in a sentence initial construction, it does not make frequent use of a passive construction, it does not have a non-referential or dummy subject constri and it dence, guage. of sur stood ment r derstar StTUCtL 2-4 SE I Ti al, Spal in Mancl this 13’ 85) def is that Specify which t u and . Cific C CUsSed I tiOHS’ l topics. Sentehci tenCe ti nition tence " 30 construction, it does have a "double subject" construction, and it does not have subject-verb agreement. From this evi— dence, Mandarin is best understood as a topic prominent lan- guage. Since English can only claim the first feature--that of surface coding of the sentence topic--it is best under- stood as a subject prominent language. Since the topic-com- ment relationship is so important in Mandarin, a better un- derstanding of how sentence topics are encoded into topic structures in Mandarin syntax is necessary. 2.4 Sentence Topic in Mandarin Syntax The definition of a sentence topic as setting a tempor- al, spatial, or individual framework works particularly well in Mandarin and other topic prominent languages. Indeed this is how both Chafe (1976:50) and Li and Thompson (1981: 85) define sentence topics for Mandarin. What this means is that the sentence topic restricts the domain, either by specifying a time, location, or individual (or object), in which the remaining comment must function. Both Chafe and Li and Thompson arrive at this definition based on the spe- cific characteristics of topic prominent languages (as dis- cussed above)--particularly the double subject construc- tions, and then determining functional uses of the sentence topics. To say that the sentence topic is simply what the sentence is about does not adequately define Mandarin sen- tence topics. What is needed is this more specified defi- nition of creating a framework in which the remaining sen- tence will hold. The fir: MlSters Set the entitie charactE is overg tenCe tc initial PhraSe’ at both SEtS the 31 Therefore, the two most important keys to determining a Mandarin sentence topic are determining the sentence initial phrase (explicit or implied), and determining if that phrase functions to set a spatial, temporal, or individual frame- work for the rest of the sentence. Clearly, such things as conjunctions and vocatives, though they occur sentence ini- tially, do not set any framework for the remainder of the sentence. 15. Yanli text, line 12 "Er wei xiansheng nimen shuo de zhei xie two CLSF mister 2pl say RLV this CLSF zi, dou xie zai shenme dongxi character all write at what thing shangbianr a? on PRTL "Two misters, what are the characters, that you are talking about, written on?" The first phrase in this sentence is er wei xiansheng 'two misters.’ But this is a vocative, used by the speaker to get the attention of the two people he is addressing. The entities which‘really set the individual framework are the characters that the two men are discussing. Therefore, it is overgeneralizing a bit too much to say that every sen- tence topic in Mandarin will be encoded in the sentence initial phrase. But if it is not encoded as the first phrase, it will certainly be encoded soon after. By looking at both sentence initial phrases and by determining what sets the framework for the sentence, Mandarin sentence topics confide 2. which t comment 32 topics can be identified with at least a fair amount of confidence. 2.4.1 Restricting the Spatial Framework. One way in which the sentence topic can specify a domain in which the comment holds is by restricting the location. 16. Da Hui Lang text, line 61 Zhe ge da guizi limian ya, zhe ge da this CLSF big cabinet inside PRTL this CLSF big hui lang bu hui qu kai guizi. grey wolf no will go open cabinet The big grey wolf will not open the cabinet. Line 61 of the Da Hui Lang text uses the noun phrase zhe ge da guizi limian ya 'inside the big cabinet’ as the topic structure encoding the sentence topic. It specifies the exact location where the wolf will not look, and establishes the cabinet as the reference point within which the comment will apply. Though this particular sentence topic does not have a wider domain than this specific line, quite often this type of sentence topic also establishes the overall setting for the remainder of the episode, and quite possibly the text. But in any event, the sentence topic zhe ge da guizi limian ya identifies the cabinet as the place where the action of the following comment will occur--the spatial framework. 2.4.2 Restricting the Temporal Framework. A sentence topic can also restrict the time in which the predicate must operate in. The code spe< begi sent func fran the 8“1131 is e that dUal foil In 8C tOpiC 33 17. Dushuren text, line 5 Zai zhe ge shihou jiyu yao zou . . . at this CLSF time hurry want leave At this time, they began to hurry and wanted to leave . . . The sentence topic in line five of the Dushuren text is en- coded in the phrase zai zhe ge shihou ’at this time.’ This specifies a definite time in which the two main characters begin to hurry, and restricts the interpretation of this sentence to this particular point in time. 2.4.3 Restricting the Individual Framework. The final function of the sentence topic is to restrict the individual framework. This use is probably the one that is most like the use of sentence topics in English. In the following ex- ample from line 7 of the Dushuren text, the sentence topic is encoded by the phrase teibei na ge xiaohai ‘especially that child.’ By restricting sentence topic to one indivi— dual, the discourse participants know which character the following information will concern. 18. Dushuren text, line 7 Teibei na ge xiaohai, ta you bei le especially that CLSF child ' 33 also carry PRF shu . . . book Especially that child, he was also carrying books. 2.4.4 Other Features of Sentence Topic in Mandarin. In addition to the prototypical characteristics of sentence topics and the more specific function of specifying a spati pics tence These is usc sentei topic . even r be mor tences finall ker. Presen toPic Senten and em identi is Onl SentEn Senten not On ta is Senten is no both t 34 spatial, temporal, or individual framework for sentence to- pics in Mandarin, there are some additional features of sen- tence topics in Mandarin syntax that need to be specified. These relate more to the actual topic structure and how it is used within the sentence. Specifically, in Mandarin, a sentence does not always need to have both the subject and topic explicitly encoded as a phrase. Sometimes there is iieven no need for an explicit sentence topic. There may also be more than one sentence topic in complex and compound sen- tences which is encoded in a topic structure for each. And, finally, the topic structure may be marked by a topic mar- ker. 2.4.4.1 Sentence Topics and Subjects. Despite the presence of "double subject" sentences, in Which both a topic structure construction and a subject occur, not every sentence in Mandarin will specify both the sentence topic and subject. When the sentence topic and the subject are identical--referring to the same entity--quite often there is only one mention of that entity referred to by both the sentence topic and the subject. In the example below, the sentence topic is encoded in the noun phrase ta ‘he.’ This not only sets the individual framework for the sentence, but ta is also the subject of the sentence. Since both the sentence topic andthe subject have the same referent, there is no need to redundantly repeat it--ta will function as both the encoded sentence topic, and the subject. 2 text i the se constr mands. have a In lin the tw. Howeye ture, l are thi Seven J toPic. topic , this Cc anaiYZE imPOrte and the 35 19. Da Hui Lang text, line 76 Ta daochu zhao. 3s everywhere look He looked everywhere. 2.4.4.2 Absence of Sentence Topic. Sometimes the con- text is so clear, it is not necessary to explicitly encode the sentenCe topic into the topic structure. These types of constructions are typically answers to questions or com- mands. Occasionally, simple declarative sentences will not have an explicit sentence topic. 20. Yongshi text, line 7 Jiran shi yonghsi, hai pa teng ma?! Since is brave man also fear pain INTR Since we are brave men, do we also fear pain?! In line seven in the Yongshi text the sentence topic is the two brave men (the main characters in the story). However, it is not explicitly encoded in a topic struc- ture, because in the preceding context the two brave men are the main focus of attention. Therefore, in line. seven there is no need to mention them as the sentence topic. The potential for confusing the implied sentence topic with some other entity is very low. Despite this, this construction was not very common in the nine texts analyzed for this thesis. This is probably due to the importance attached to the sentence topic in Mandarin, and the desire for it not to be mistaken for some other possible sentence topic. Sen‘ ten: ple: oft: say: 198? In j is ( Shoi fran Th1S pOit the Toll fore 36 2.4.4.3 Second Sentence Topic in Complex and Compound Sentences. Though most sentences will only have one sen- tence topic for the entire sentence, occasionally in com- plex sentences, a second sentence topic will occur. Quite often, this second sentence topic will follow "verbs of saying . . . and verbs of mental activity" (Li and Thompson 1981:100). 21. Yanli text, line Diyi ge jinshiyan . . . shuo: "Zhe bu shi first CLSF shortsighted say this no is xiezhe 'guangming zhengzhi’ si ge da write guangming zhengzhi four CLSF big '" zi ma characters INTR The first shortsighted man . . . said: "Aren’t the characters 'guangming zhengzhi’ written in big characters on this board!" In line seven of the Yanli text, the first sentence topic is encoded in the noun phrase diyi ge jinshiyan 'the first shortsighted man.’ This character restricts the individual framework for the entire sentence. Following the verb shou 'to say,’ a second sentence topic is encoded as zhe ‘this.’ This refers to the board which the shortsighted man is pointing at, and it restricts the individual framework of the utterance of the first shortsighted man. Similarly, a second sentence topic may also occur in a compound sentence. Should the topical framework change following non-conjoining conjunctions--like suoyi ‘there- fore’ or buguo 'but’--another sentence topic construction will C the fi the nc danshi grandm phrase StruCt of sev 1981:8. ies, 01 fairly Here, 1: iiih th This is 37 will occur after the conjunction. In the following example, the first sentence topic is the old grandmother, encoded by the noun phrase wo 'I.’ But following the conjunction danshi 'but,’ the individual framework changes from the old grandmother to the two little girls, encoded by the noun phrase nimen ’you.’ 22. Da Hui Lang text, line 97 . . . "Wo keyi dai nimen xiao shan, danshi ls can take 2pl down mountain but nimen yiding de hui lai . . . 2pl certainly ADV return come "I can take you down the mountain, but you cer- tainly must come back . . . 2.4.4.4 Optional Markers. Occasionally, the topic structure in Mandarin may be immediately followed by one of several topic markers--a, ne, me, ba (Li and Thompson 1981:85). Of the three Mandarin speakers who related stor- ies, only one used topic structure markers-~ne--and then. fairly infrequently. 23. Da Hui Lang text, line 16 Liang ge nuer ne, xinggao-cailie de zai two CLSF daughter PRTL jubilant MNR DUR zhai hua. pick flower The two daughters were excitedly picking flowers. Here, the sentence topic is encoded in the topic structure with the noun phrase liang ge nuer ‘the two daughters.’ This is immediately followed by the particle ne, which marks the t: it moa the t6 above restri every seems four t tion a 38 the topic structure. As this marker is used in this text, it most commonly follows a sentence topic which restricts the temporal framework of the sentence. However, as the above example shows, it can also mark a sentence topic which restricts an individual framework. Nor is it the case that every temporal sentence topic is marked with me. Those this seems like a random patterning, it will be shown in chapter four that this use of ne, for this speaker, appears to func- tion as an episode boundary marker. 2.5 Summary The preceding chapter was concerned with specifying in greater detail the prototypical characteristics of sen- tence topics. It also addressed why Mandarin is considered- a topic prominent language, and ended with a discussion on how to identify sentence topics in Mandarin. Six prototypical characteristics of sentence topics were listed. This included the discourse function of sen- tence topics linking the sentence in which the sentence topic occurs to the rest of the discourse. Another charac- teristic is the predominance of sentence topics occurring in sentence initial position. Sentence topics must also be part of the pragmatic or presupposed knowledge of the participants. They will also be encoded in definite phrases, and either have an explicit earlier mention in the discourse or be inferrable from other information in the discourse. eXplic senten Show h tempor Mandari tioned Senteh Option iamili. prinCe ARd it Handar: 39 Mandarin is considered a topic prominent language because of certain syntactic features common to other topic prominent languages but not to subject prominent languages like English. Specifically, Mandarin has a specific topic structure construction to encode the sentence topic. Mandarin also makes little use of the passive construction, nor does itemploy dummy subjects. Mandarin does have dou- ble subject constructions, but lacks subject-verb agreement. Finally, the use of Mandarin sentence topics was more explicitly stated as a means to better be able to identify sentence topics in Mandarin. Examples were provided which show how Mandarin uses sentence topics to set a spatial, temporal, or individual framework. And other features of Mandarin syntax as they relate to sentence topics were men- tioned: sentence topic and subject use, lack of an explicit sentence topic, use in compound and complex sentences, and optional sentence topic markers. The next chapter will discuss Prince’s model of assumed familiarity. ‘As such it will discuss the distinctions Prince makes among Brand New, Inferred, and Evoked entities. And it will be shown how this taxonomy can be applied to Mandarin sentence topics. PRINCE’S TAXONOMY OF ASSUMED FAMILIARITY 3.1 Introduction Based on the above discussion, it is now possible to take a Mandarin text and determine what the sentence topic is for each line. But a list of sentence topics is hardly interesting or insightful in and of itself. A still better understanding of Mandarin discourse structure can be had if this list of sentence topics could be analyzed to deter- mine what type of informational entities typically function as sentence topics. If a sentence topic must be presupposed information, how is this achieved by the narrator? There are a variety of models proposed to answer this question, but of these, Ellen Prince’s (1981) is one of the more de— tailed._ The following sections will describe her taxonomy of informational entities--new, evoked, and inferred. After this will be a section describing how Prince applied this taxonomy to analyze English noun phrases. The next section will discuss how this taxonomy can be used for Mandarin sen- tence topics. The final section will summarize the material presented in this chapter. 3.2 Prince’s Taxonomy of Given-New Information Prince devised her taxonomy of given-new information because, though it was a common distinction made by many 40 li te th 3P of ab be ta km im it Dre Thi 41 linguists in a variety of fields, no one model to charac- terize this distinction was completely adequate or used in a consistent manner (Prince 1981:225). Her model describes three different levels of shared knowledge between the speaker and the other discourse participants. These levels of shared knowledge are new, evoked, and inferred, each of which have their own subgroupings, which will be discussed below. The following is a diagramatic model of Prince’s taxonomy of shared knowledge or assumed familiarity. Assumed Familiarity ‘-“‘~“§~§§‘~““-~a //;New Infer{:: Evoked /BN\ U I . 10 E ES BN BNA BN Brand New (unanchored) BNA Brand New Anchored U Unused I Inferrable (non-containing) Containing Inferrable E Textually Evoked ES Situationally Evoked 3.2.1 New Information. The first level of shared knowledge is new information. This is information that is introduced for the first time into the discourse. As such, it is new information, representing information that was not previously perceived as being an element of the discourse. This new information can be one of two types--brand new or unused. 42 3.2.1.1 Brand New Information. Brand new information is precisely that--it is information that is completely new to the discourse. 1. Si Maguang text, line 9 Ta zhao le yi zhao, zhaodao yi kuai shitou, 3s look PRF one look find one piece stone yi kuai da shitou. one piece big stone After looking around for awhile, he found a large stone. - In line nine of the Si Maguang text, the large stone repre- sents brand new information in the discourse. It has no previous mention in the text, and the speaker would not assume that the listeners would presuppose this information. Brand new information can also be furthered divided into two subcategories: ’anchored and unanchored. Anchored information is "linked by means of another NP, or 'Anchor,’ properly contained in it, to some other discourse entity" (Prince 1981:236). Therefore, the noun phrase 'a book I read’ anchors brand new information, 'a book,’ with another noun phrase 'I.’ The anchor, itself, must, however, be information that already has been established into the text. To use an anchor that is also brand new information results in an odd sentence. 2. Yesterday, I talked to a guy I know from work. 3. ?Yesterday, I talked to a guy a man knows from work. tnanc matic in yi diffe posed there the l Rathe the c COUI‘S In th. or um' ata are a Only as in Howev Schoo knOw . preSe: But t] iOUSl: ibOut 43 Unanchored brand new information is simply brand new infor- mation that is introduced without the use of an anchor--as in yi quai da shitou in example one. 3.2.1.2 Unused Information. Unused information is different from brand new information in that it is presup- posed information. The first mention of the unused entity, therefore, does not create a new piece of information which the listener must add to his or her knowledge of the text. Rather, an unused entity simply brings this information to the conscious awareness of the listener. Proper names, whiCh would be recognized by all dis- course participants, are examples of unused information. 4. Talmy Givon teaches at the University of Oregon. In the above example, Talmy Givon can either be brand new or unused information. If the speaker is addressing people at a linguistic’s conference, chances are that those people are aware of who Talmy Givon is. It is new information only in the sense that the listeners will now register Givon as information that is currently important to the discourse. However, if the same utterance was spoken during a Sunday school class, it is unlikely that very few individuals would know who Talmy Givon is. This information, then, will re- present brand new information for this group of listeners. But this use of brand new information can only be felici- tously employed if the speaker wanted to make some point about Mr. Givon relevant to the class. iarit "alre to sea aware textu entit In 1i: added Phras. in 11: tion.. a pPe~ extra. Settil the Pa in Wh- 44 3.2.2 Evoked Information. The second level of famil- iarity is evoked information. This is information that is "already in the discourse model" (Prince 1981:236). That is to say, this is information that the listeners are already aware of. Entities can be evoked in one of two ways--either textually or Situationally. Textually evoked entities are entities which have been mentioned earlier in the text. 5. Dushuren text, line 3 Ta dai 1e hen duo shu, yong shengzi kun hao 33 take PRF very many book use rope tie well 1e . . . PRF He had many books, which were tied together with a rope . . . 6. Dushuren text, line 11 Shengzi duan le. rope break PRF . The rope broke. In line 3 of the Dushuren text, brand new information is added to the discourse--that of a rope encoded by the noun phrase shengzi. Therefore, the next mention of the rope in line 11 of the text represents textually evoked informa- tion--information that the listeners are aware of based on a previous explicit mention of the entity. Entities can also be Situationally evoked through the extratextual context of the discourse. This refers to the setting of the discourse, which would include such things as the participants in the discourse and the physical location in which the discourse occurs. In the following example, ‘n 1 assun is ha tione In th situa with to in ties and s iarit the 1 Other 45 assume the extratextual context contains a picture which is hanging over a couch--neither of which have been men- tioned yet in the discourse. 7. I think that painting over the couch is nice. In this example, there are three entities which have been Situationally evoked. The first is the speaker, encoded with the pronoun 'I.’ The second is the painting referred to in the utterance, and the third is the couch. All enti- ties are part of the extratextual context of the discourse, and so the participants would be aware of them. 3.2.3 Inferred Information. The final level of famil- iarity is inferred information. This is information that the listeners can deduce through logic or common sense from other information presented in the discourse. 8. Da Hui Lang text, line 1 You yi jiaren, ta jia 1i you si ge have one family 33 family in have four CLSF nuer. daughter There was a family which had four daughters. 9. Da Hui Lang text, line 103 Ta de liang ge da jiejie jiu jia 1e 33 GEN two CLSF big older sister then marry PRF nanren gen tamen de nanren zou le. man with 3pl GEN man leave PRF Their two older sisters then married men and with them left. In t} provi there the 5 refer 'the menti! ble f Chara text. entit 46 In the first line of the Da Hui Lang text, the listeners are provided with information about a family, specifically that there are four daughters in this family. At the very end of the story, in line 103, two older daughters are explicitly referred to with the noun phrase ta de liang ge da jiejie 'the two older sisters.’ Although there has not been any mention of these two characters earlier, it is still possi- ble for the listeners to infer the identity of the two characters from the information presented in line one of the text. Inferred entities can either be noncontaining, like the ~ example above, or containing. A containing inferrable is defined by Prince as one in which "what is inferenced off of is properly contained within the inferrable NP itself" (Prince 1981:236). Instead of the inference relying on information presented in a previous line, the inference happens within the one noun phrase in which the inferrable entity is encoded. 10. In the Fall, the leaves of my parent’s trees turn bright red and orange. In this example, 'the leaves of my parent’s trees’ encodes information through a containing inferrable. By a logi- cal relationship of part to whole, it is possible to infer the identity of the leaves from the trees. All of the in- formation that is necessary to make this inference is en- coded by the noun phrase 'the leaves of my parent’s trees.’ _.____——‘ 3.3 plai: text forme that notec branc tages (Prim oral Engli tende evoke with Ven,’ terms numbe] narra1 thePe down j Chafe Opposfi Jeets exact 1n man 47 3.3 Applying the Assumed Familiarity Taxonomy to English Using her taxonomy of assumed familiarity, as ex- plained above, Prince went on to analyze an English oral text to determine the patterns of usage of the various in- formational entity types. Very briefly, her analysis showed that of all the grammatical_subjects in her text, 93.4% de- noted evoked entities, 6.6% denoted inferred, and 0% denoted brand new. For nonsubjects, on the other hand, the percent- tages were 48.8% evoked, 30.2% inferred, and 20.9% brand new (Prince 1981:243). Since this analysis is based on only one oral text, no firm conclusions can be applied in general to English, but her analysis does suggest the overwhelming tendency for subjects in oral English narratives to encode evoked entities. It is interesting to note that Chafe (1987) working with a slightly different model (employing the terms 'gi- ven,’ 'new,’ and 'accessible’ instead of Prince’s respective terms 'evoked,’ 'new,’ and 'inferred’) reaches very similar numbers. He states, "of the 23 starting points of the narrative, 20 were given, three . . . were accessible and there were none that were\new" (Chafe 1987:37). This breaks down into 86.9% given, 15% accessible, and 0% new. However, Chafe was analyzing starting points (his term for topics) as opposed to subjects in English. Even though in English sub- jects and topics almost always coincide, there cannot be an exact comparison of Chafe’s findings to Prince’s. And yet, in many ways it does support her analysis. . _‘.q 3.4 ity. were this ever: sents frame tione 8 prc evoka trodu 48 3.4 Prince’s Taxonomy and Mandarin Sentence Topics When Prince devised her taxonomy of assumed familiar- ity, she did so to better characterize those entities which were encoded by noun phrases. One difficulty with applying this taxonomy to Mandarin sentence topics, then, is that not every sentence topic is encoded by a noun phrase. Those sentence topics which are used to set a temporal or locative framework are quite often encoded in adverbial and preposi- tional phrases, respectively. But this does not have to be a problem. Both time and location can represent new, evoked, or inferred information. In the following example, a locational setting is in- troduced as new information, and then later evoked. 11. Da Hui Lang text, line 6 Ranhou ba tamen ren zai shan shang. then DO 3pl leave at mountain on Then, he will leave them on the mountain. 12. Da Hui Lang'text, line 8 Shan shang you hen duo xian hua. mountain on have very many fresh flower On the mountain there are many wild flowers. Line six of the Da Hui Lang text introduces new information, the setting of the mountain, encoded in a prepositional phrase. Later, in line eight, this same setting, again en- coded as a prepositional phrase, is evoked, and becomes the sentence topic of the sentence. It is also possible to infer settings. In the Dushuren text, the two characters are travelling to a city when the GVET. No p. stil; trave or pa the t trave throu forma entit tity. tempo 1" thi enCOdE aftEr 49 events recorded in line nine occur. 13. Dushuren text, line 9 Jieguo zai lu shang, shuai dao le. therefore at road on fall down PRF Therefore, on the road, he fell down. No previous mention of the road is made in the text. But still, it is information that can be inferred. When people travel to and from cities, it is typically done on roads or paths. So even though it is never stated explicitly in the text prior to line nine that the two characters were travelling on a road, it is information that can be inferred through common sense reasoning. Temporal information is different from locational in- formation in that temporal information refers to an abstract entity while locational information refers to a concrete en- tity. Therefore, very often, sentence topics which set a temporal framework represent inferred information. 14. Dushuren text, line 10 Yi shuai dao le yihou, ta de shu wanchuan soon fall down PRF after 33 GEN book completely san le. fall apart PRF Soon after he fell down, his books came apart. In this line from the Dushuren text, the sentence topic is encoded by the adverbial phrase yi shuai dao 1e yihou 'soon after he fell down.’ This is information which can only be inferred from information from the preceding context. Even thou refe Such SGHS' exam -— — tenet 3.5 milaz Couri infei duced entia titie has 6 the e those other into Prime the n &@1311 50 though the preceding line (see example 13) mentions that the character fell down, the sentence topic in line 10 refers to a time right after the events in line 9 happened. Such information can only be inferred through the common sense reasoning that events happen in sequence. The above example is very typical of the way in which a temporal sen- tence topic is introduced via inference. 3.5 Summary This chapter presented Prince’s taxonomy of assumed fa- milarity. This taxonomy separates the entities of a dis— course into three levels of information: new, evoked, and inferred. New entities are those which have not been intro- duced into the discourse before, and can be further differ- entiated between brand new and unused entities. Evoked en- tities are entities in which the phrase encoding the entity has either been previously used in the discourse or are in the extratextual context. And lastly, inferred entities are those which the referent of the entity can be deduced from other information in the discourse. These can be divided into either non-containing inferrables or containing in- ferrables. Following this section, was a brief section on how Prince applied this taxonomy in a Study of an oral text from English, and the conclusions she was able to at least tentatively draw from her analysis. In a similar fashion, the next section discussed how Prince’s taxonomy can be applied to Mandarin sentence topics. Even though Prince 1...“ ” desi demo sent phra topir 0mY» entit level to de with 51 designed her taxonomy in reference to noun phrases, it was demonstrated that the taxonomy can also work for Mandarin sentence topics which are not always encoded by noun phrases. The next chapter begins the actual analysis of sentence topics and oral Mandarin narratives. Using Prince’s taxon- omy, the Mandarin sentence topics will be identified as entities representing information from one of the three levels of assumed familiarity. These will then be analyzed to determine how other discourse parameters are correlated with level of familiarity. THE 4.1 I may i ther The a corre pendi this into break. in the be thi Paramfi 0f the ePiSoc tenCe, Conclu And t” Stat-l8 MUCh 0 Cal an exCapt THE FUNCTION OF SENTENCE TOPICS IN ORAL MANDARIN NARRATIVE 4.1 Introduction This chapter will examine some of the parameters which may influence the use of one level of familiarity over ano- ther in the mention of a sentence topic in oral narrative. The actual list of sentence topics from each text and its corresponding level of familiarity can be found in the Ap- pendix--only the analysis of the data will be recorded in this chapter. The remainder of this chapter will be divided into six sections. The first will present the simple breakdown of the number of each level of familiarity used in the mention of a sentence topic. Following this will be three sections, each of which will examine a separate parameter and how it influences sentence topics. The first of these will examine animacy. The second will examine episode boundaries, and the third will examine topic persis- tence. The fifth section of this chapter will draw some conclusions based on the findings of the previous sections. And the sixth will summarize the chapter. One additional note is that for the most part an actual statistical analysis will not be done on the data presented. Much of the data show clear enough patterns that a statisti— cal analysis would not add much more insight. The one exception to this will be the section on episode boundaries. 52 “'5 4.2 sent use « Prin< used situe text- ly ex esser will Isis disti 53 4.2 Levels of Familiarity and Sentence Topics in Oral Texts Before any deeper investigation into the selection of sentence topics can be done, the actual patterning of the use of each level of familiarity must be understood. Though Prince breaks her taxonomy into six levels--brand new, un- used, inferred, containing inferred, textually evoked and Situationally evoked--only four were present in any given text-~brand new, inferred, containing inferred, and textual- ly evoked. Since inferred and containing inferred represent essentially the same level--just different codings--these will be combined into one level--inferred. All of the anal- ysis to follow, therefore, will make use of this three way distinction of brand new (BN), evoked (E), and inferred (I). The breakdown of the three oral texts into the number of sentence topics used at each level of famiality is given in Table 1 below. Beside each figure is a percentage based on the number of occurrences for that level of familiarity divided by the total number of sentence topics for that particular text. Therefore, for the Da Hui Lang text, each number is divided by 140, for the Si Maguang text 13, and for the Dushuren text 13. The percentages for the combined figures is based upon a total of 166 sentence topics. Table 1: Oral Texts—-Levels of Familiarity BN E I Da Hui Lang 1 (0.7%) 106 (75.7%) 33 (23.6%) Si Maguang 1 (7.7%) 10 (76.9%) 2 (15.4%) Dushuren 0 (0.0%) 8 (61.5%) 5 (38.5%) Combined 2 (1.2%) 124 (74.7%) 40 (24.1%) thre topi rare infe time real. The a ferre time first tOpiC would other famil will 4.3 it me: m&Cy : Synta( affect rin. constr human one Of 54 Table 1 presents the simple information that for all three texts, there is a much greater use of evoked sentence topics than the other two. Brand new sentence topics are rarely used (not at all in fact in the Dushuren text), and inferred sentence topics a little under one fourth of the time. Though this is an interesting pattern, what does it really mean? Actually, this information really does not say much. The above figures, taken by themselves, suggest that an in- ferred sentence topic, for example, will occur 24.1% of the time in an oral text. This conceivably could mean that the first quarter of the text could be one inferred sentence topic after another, while the remaining three quarters would all be evoked. But this is not the case at all, for others factors will influence the selection of one level of familiarity over another, as the following three sections will illustrate. 4.3 Animacy and Sentence Topic in Oral Texts The first parameter which will be examined to see how it may or may not affect sentence topics is animacy. Ani- macy is often an important feature in the coding of various syntactic structures (Croft 1990:127-130). Certainly it affects the encoding of the genitive construction in Manda- rin. Therefore, if both the noun phrases in the genitive construction have a high degree of animacy (a pronoun and a human noun) the genitive marker de is not required. But if one of the two noun phrases in the genitive construction has a 10 This cons stru: of se 55 a low degree of animacy the genitive marker is required. This means that wo meimei 'my younger sister’ is a correct construction, but wo mao 'my cat’ is not the correct con- struction--wo de mao is. But how does this affect the use of sentence topics? I Table 2 below gives the breakdown of the number of sen- tence topics used at each level of familiarity as it relates to an animacy scale. For the purposes of this paper, ani- macy has been divided into three levels: human (H), inan- imate (N), and abstract (A). Human, obviously, is used for all human noun phrases, but it also is applied to the wolf in the Da Hui Lang text. Since the wolf talks, thinks, and functions as a human, it should be regarded as having the same level of animacy as do the old grandmother and the two little girls. Inanimate refers to all concrete nouns that are not human--ropes, roads, mountains. Abstract is used for abstract concepts such as time and human charac- teristics (like age). Oddly enough, there were no animals, apart from the humanized wolf, in any of texts, oral or written. The right hand column of Table 2 gives the individual percentage for each level of animacy as it occurs within each level of familiarity. This is based upon the total number of occurrences of a level of animacy divided by the total number of sentence topics which occurred at a parti- cular level of familiarity. For example, the total number of Brand New sentence topics is two. Of these, both refer ‘1 to h sent of s Lang text Tabl The Very entit tenCe entit Stron which it re Sente. which hUman 56 to human entities, which means that 100% of all Brand New sentence topics will have human referents. For the sake of space, each text has been abbreviated: L for the Da Hui Lang text, M for the Si Maguang text, and D for the Dushuren text. Table 2: Oral Texts-~Animacy and Sentence Topic L M D Occurrence Total Percent H 1 1 0 2 2 100.0% BN N 0 0 0 0 2 0.0% A 0 0 0 0 2 0.0% H 91 9 5 105 124 83.9% E N 14 1 2 17 124 13.7% A 1 0 1 2 124 1.6% H 3 0 0 3 40 7.5% I N 0 0 1 1 40 2.5% A 30 2 4 36 40 90.0% Several observations can be made based upon Table 2. The first is that Brand New sentence topics, which have a very low occurrence in these three texts, are only used with entities with a human level of animacy. Additionally, sen- tence topics, which are evoked, are strongly tied to human entities. And Inferred sentence topics are even more strongly linked to abstract entities. These patterns can be seen more clearly in Table 3, which compares the occurrence of the level of familiarity as it relates to the animacy scale. So that all the human sentence topics, for example, are grouped together to see which level of familiarity is most often used to refer to a human entity. 57 Table 3: Oral Texts-—Animacy and Levels of Familiarity Total BN E I Human 110 2 (1.8%) 105 (95.5%) 3 (2.7%) Inanimate 18 0 (0.0%) 17 (94.4%) 1, (5.6%) Abstract 38 0 (0.0%) 2 (5.3%) 36 (94.7%) From this table, it is very clear that a human entity is 95.5% more likely to be referred to with an Evoked sentence topic. Inanimate entities are also just as likely to be re- ferred to with an Evoked sentence topics. And finally, ab- stract entities are highly probable to be referred to with Inferred sentence.topics. Before any hypotheses can be made as to why these probabilities exist, other parameters should be examined to see how they affect sentence topics. 4.4 Episode Boundaries and Sentence Topic in Oral Texts In addition to animacy, the Choice of one level of information over another in the use of sentence topics may be influenced by the larger discourse structure. Anyone who has heard or read a story will know that the story can of- ten be divided into smaller units or episodes--specific sec- tions which focus on one character or one event. It is en- tirely likely that one level of familiarity is most often associated with the first line of a new episode. The first key to discovering the possibility of this correlation is to know what an episode is. Tomlin (1987: 461) gives a very simple diagnostic in determining episode boundaries--"major changes in time, place, or characters correspond to episode boundaries." This very neatly ties into the various uses for sentence topics in Mandarin-- 58 that of setting a temporal, locational, or individual frame- work. Therefore, any of the three types of Mandarin sen- tence topics can be used to signal an episode boundary. Note that although an episode boundary denotes a change from one context to another--from an old setting to a new one--this does not mean that the sentence topic signalling the episode boundary represents new information, and there- fore cannot be a sentence topics as defined in this thesis. For example, in the Yanli text, the information that there are two shortsighted men in this Story is introduced early, and hence represent old-given information. One episode cen- ters on the first shortsighted man and what he believes is written on an inscribed board. The episode which follows changes the focus from the first shortsighted man to the second one. Although this represents a new episode, the information about the second shortsighted man is still old- given information. The change in episode boundary may re- quire the encoding of the sentence topic as a full noun phrase (as discussed in chapter 2), but this does not represent new information. Episodes, in and of themselves, can have different functions within the text. Several linguists have sought to divide up the structure of a narrative in a variety of ways. Labov 1972 lists these categories as follows: Abstract Orientation Complicating Action Embedded Orientation Evaluation Coda ODU'IvfiODNt-e 59 Of these, four are the most pertinent for the purpose of this paper. The first is the 'Orientation.’ This sets the stage for the remainder of the discourse. It enables the participants to get at least a rough idea of the basic set- ting of the text to follow. Next is the ’Complicating Ac- tion.’ This recounts activities which occur in the text. ‘Embedded Orientation’ provides additional background infor- mation within the text itself. And the last section is the ‘Coda.’ This is typically the wrap up of the story. The following mini-narrative illustrates these four categories: Orientation: Once upon a time in a kingdom far, far away lived a beautiful princess. Complicating Action: One day, she found a toad and impetuously kissed it. Embedded Orientation: The toad was actually a handsome prince, who was under an evil spell which only a kiss would break. Coda: The princess and the prince got married and lived happily ever after. This is not to suggest that every time a line can be classi- fied into one of these structural units that it corresponds to an episode boundary. Rather this is a useful way to be- gin to break down the structure of a text into more manage- able units so that finer divisions in episode boundaries can be made. For example, a cluster of sentences which are all complicating actions could indicate an episodic unit. 60 This is not the only way to divide up a text. Longacre 1983 does not make the fine distinctions as proposed by Labov. He divides a text into two basic components-- background and foreground. Background typically consists of descriptive comments, much like the ‘Orientation’ and 'Em- bedded Orientation’ above. Foreground represents important activities in the text, like 'Complicating Action.’ But Longacre does include a division in his model which is left out in Labov’s. This is the notion of ‘Peak.’ The peak of a story is point of highest tension, such as when the wolf is searching for the two little girls in the Da Hui Lang text. To reflect this tension, the peak is often quite distinct from other episodes in the text. Changes in the tense/aspect system may occur and other prototypical dis- course features may be absent (Longacre 1983:25). These peak marking features will often continue to occur until the resolution of the tension is resolved. This is readily apparent in the Da Hui Lang text. This was the only text to make use of the optional sentence topic marker ne. At first, the distribution of ne in this text seemed arbitrary. But when the narrative was broken down into episodes, it became very apparent that at every episode boundary, ne was used to mark the sentence topic. However, ne is not used throughout the entire text. There is a sec- tion from line 63 to 99 in which no me is used, but this same section is clearly not one prolonged episode.) What it is, though, is the peak (from line 63 to 85) and the 61 resolution (lines 86-98). Since this particular narrator . was the only one to even use ne, it is not possible to even suggest that ne is an episode marker in Mandarin narrative. And yet, for this one speaker, it serves this function. Clearly, a detailed analysis of the discourse struc- ture of each text was not possible, or even desirable. But based on the criteria above, each text has been divided into episodes. These episode divisions are reCOrded with their respective texts in the Appendix. What follows are the correlations found between episode boundaries and the level of information used at these boundaries in the mention of the sentence topics. Table 4 below gives the number of episode boundaries which occurred for each text as they related to the levels of information. In the right column are the percentages of the occurrence in one level of information with the total possible number of episode boundaries (29). Table 4: Oral Texts-—Episode Boundaries and Levels of Familiarity Lang Maguang Dushuren Total Percentage BN 1 1 0 2 6.9% E 3 3 2 8 27.4% I 15 1 3 19 65.5% Table 4 clearly shows the preponderance of Inferred sentence topics occurring at episode boundaries. Inferred sentence topics are almost three times as likely as Evoked sentence topics to be used in this way. And Brand New sentence to-, pics appear to be used hardly at all. 62 Another way to look at this information is to compare the total number of occurrences of each level of familiarity with the total number of times that level was used at an episode boundary. Table 5: Oral Texts--Episode Boundaries and Total Occur- rences of Levels of Familiarity Episode Boundaries/Occurrences Percentages BN 2 /2 100.0% E 8/124 6.5% I 19/40 47.5% Based on Table 5, it is now Brand New sentence topics with the greatest occurrence at episode boundaries. This only makes sense, since both uses of Brand New sentence topics occurred in the very first line of the narrative, and the first line has to be an episode boundary. This table also shows that Evoked sentence topics are very infrequently used at episode boundaries, while almost half of all the Inferred sentence topics are used at episode boundaries. Even greater insight can be achieved by looking at how animacy correlates with episode boundaries. Table 6: Oral Texts-~Episode Boundaries and Animacy Episode Boundary Total Occurrences Percentage H 2 2 100.0% BN N 0 0 0.0% A 0 ‘ 0 0.0% H 7 105 6.7% E N 0 17 0.0% A 1 2 50.0% H 0 3 0.0% I N 0 1 0.0% A 19 36 52.8% 63 From this table some further observations can be drawn. Obviously, since Brand New sentence topics are always human entities, Brand New human sentence topics will always occur at a sentence boundary. Additionally, no inanimate sentence topic is ever used at an episode boundary. However, ab— stract entities occur at least 50% of the time at an episode boundary no matter if the sentence topic is evoked or in- ferred. Here, it could be useful to run a statistical analysis on the above data in an attempt to tease out any further in- formation which may be relevant. To do this the Varbrule statistical program was used. Since Brand New sentence to- pics only occur at episode boundaries with human entities, and inanimate sentence topics never occur at episode bound- aries, only two levels of familiarity--evoked and inferred-- and two levels of animacy——human and abstract--were analyzed to determine the probabilities of occurrence at episode boundaries. The results of the analysis showed pretty much what was expected. Table 7 gives the specific probabilities for each factor. Table 7: Oral Texts--Probability and Episode Boundaries Probability Evoked 0.503 Inferred 0.497 Human 0.198 Abstract 0.802 What this means is abstract entities strongly promote the use of an episode boundary, human entities not at all, and 64 Evoked and Inferred sentence topics are pretty much neutral in the promoting or retarding of episode boundaries. The Varbrule program also allows a stepwise regression which identifies factors which are not significant for the parameter being examined--in this case the occurrence of an episode boundary. In this case the stepwise regression showed that level of familiarity is not significant while level of animacy is. This is to be expected from the near equal percentage of abstract entities at episode boundaries regardless of the level of familiarity of the sentence topic. However, it should be remembered that the majority of Inferred sentence topics are at episode boundaries, while the vast majority of Evoked sentence topics are not at epi- sode boundaries. 4.5 Topic Persistence and Sentence Topics in Oral Texts A final parameter to be investigated that may influence the use of sentence topics is topic persistence. Specifi- cally, do entities with a relatively high topic persistence occur with a certain level of familiarity, while those with low topic persistence occur with another level of familiar— ity? Topic persistence is a cataphoric measure of the num- ber of following clauses which include a mention of the topic being counted (Givon 1990:908). There are two ways Suggested for doing this. One is to count the number of clauses in which the referent of the topic is mentioned (including zero anaphor) in the succeeding ten clauses. The other is to just count the number of consecutive clauses in 65 which the referent is mentioned. For this paper, the first method was adopted. A topic persistence count for the sentence topic Nanguo xiansheng 'Mr. Nanguo’ from the Duzou text, lines 2 to 4, illustrates how this is done (clause boundaries are indi- cated by a slash): You ge Nanguo xiansheng/‘¢ zhidao zhei -ge have CLSF Nanguo mister know this CLSF qingkuang/ ¢ jiu jian Xuanwang,/ ¢.shuo/ ziji chui de situation then meet Xuanwang say self play MNR ruhe ruhe/ ¢ qingqiu/ ¢ canjia zhei ge yuedui/ how how please ask join this CLSF orchestra ¢ wei Xuanwang chui yu./ Xuanwang ba ta bianjin for Xuanwang play reed pipe Xuanwang DO 33 put yuedei/ binqie gei ta hen gao de xinshui./ orchestra moreover give 3s very high PRTL salary Xuanwang si le,/ Minwang jie wei./ Xuanwang die PRF Minwang approach place There was a Mr. Nanguo, who knew of this situation, went to see Xuanwang, and said that he knew how to play. He asked if he could join the orchestra and play the reed pipe for Xuanwang. Xuanwang put him in the orchestra and gave him a high salary. Xuan- wang died, and Minwang took the throne. In the ten clauses following the first mention of Nanguo xiansheng, he is referred to eight times. Therefore the to- pic persistence count of Nanguo xiansheng is eight. This is then averaged with other scores to arrive at an overall to- pic persistence score. The higher the topic persistence score, "the more it (the entity) recurs in the following discourse and the more topical it is" (Myhill 1992: ch. 2, pg. 15), and the more important it is to the discourse. 66 The topic persistence score was calculated for each sentence topic in the three oral texts. Table 7 gives the total topic persistence score for each text followed by the total number of sentence topics which occurred at the perti- nent level of familiarity. Therefore, in the Da Hui Lang text, for brand new entities, the topic persistence score was four. This is divided by the number of occurrences of brand new sentence topics in the text, which was one. The individual scores from each text are added together for each level of familiarity to arrive at an overall topic persis-l tence score (TP). Table 8: Oral Texts--TP and Levels of Familiarity Lang Maguang Dushuren Total TP BN 4/1 1/1 0/0 5/2 2.50 E 445/106 16/10 29/8 490/124 3.95 I 9/33 0/2 . 0/5 9/40 0.23 Table 8 shows that of the three levels of familiarity, Evoked sentence topics have the highest topic persistence, followed by Brand New sentence topics, and finally Inferred sentence topics, which have an extremely low score. Just as with episode boundaries, further distinctions can be made by comparing topic persistence with animacy. Table 9 presents the same information as Table 8 above, except Table 9 breaks down each level of familiarity into the three levels of animacy. 67 Table 9: Oral Texts--Animacy and Topic Persistence Lang Maguang Dushuren Total TP H 4/1 1/1 0/0 5/2 2.50 BN N 0/0- 0/0 0/0 0/0 0.00 A 0/0 0/0 0/0 0/0 0.00 H 420/91 16/9 27/5 463/105 4.41 E N 25/14 0/1 2/2 27/17 1.59 A 0/1 0/0 0/1 0/2 0.00 H 6/3 0/0 0/0 6/3 2.00 I N 0/0 0/0 0/1 0/1 0.00 A 1/30 0/2 0/4 1/36 0.03 From this it is possible to derive a ranking of sentence topic types from highest to lowest in topic persistence. Evoked Human TP = 4.41 Brand New Human TP = 2.50 Inferred Human TP = 2.00 Evoked Inanimate TP = 1.59 Inferred Abstract TP = 0.03 Evoked Abstract TP = 0.00 Inferred Inanimate TP = 0.00 Brand New Inanimate TP = 0.00 (no occurrence) Brand New Abstract TP = 0.00 (no occurrence) This hierarchy clearly indicates that human entities are more "topical"--that is these entities are more salient and more important to the narrative--than the other levels of animacy. Next in the hierarchy are evoked inanimate entities, and finally Inferred abstract entities. All of this makes sense, since the characters (the human entities) are the most important aspect of a story, followed by various props and settings (the inanimate entities) which are used by the characters, and finally the marking of time and other less tangible things (the abstract entities). Note that Evoked abstract and Inferred inanimate entities 68 not only have a topic persistence score of zero, but they also are used very rarely in the texts--suggesting that they represent entities which are not very important to the over- all narrative. 4.6 Discussion What does all of the above information have to say about the use of sentence topics in oral Mandarin narrative? At the beginning of this chapter, it was noted that of the three levels of familiarity, sentence topics were evoked 74.7% of the time, inferred 24.1% of the time, and brand new 1.2% of the time. Now that other parameters, like animacy, episode boundaries, and topic persistence, have been exam- ined, a more informed explanation can be given for the use of sentence topics as noted above. 4.6.1 Brand New Sentence Topics. Despite the very low usage of Brand New sentence topics, they have a rather im- portant function. Occurring text initially, they provide the orientation for the narrative by introducing somewhat important characters into the text--thus their exclusive use of human entities and their relatively high topic per- sistence. Brand New sentence topics have such a low occur- rence, then, not because they contain unimportant informa- tion, but rather because once the discourse is initiated evoked and inferred information can be used. Therefore, they function to start the narrative. 69 4.6.2 Evoked Sentence Topics. Evoked sentence topics have the highest percentage of usage of the three levels of familiarity. They are also almost exclusively used for hu- man entities (95.5%) and most inanimate entities (94.4%), as well as having the highest topic persistence (3.95). Despite this, they are used only about a quarter of the time for episode boundaries, and only 6.5% of all Evoked sentence I .. A—JI‘." topics actually occur at an episode boundary. Basically, then, Evoked sentence topics function to topicalize the characters of the narrative. Since a narra- tive is essentially about characters--who they are and what they did--the great preponderance of sentence topics will naturally concern the characters. The clearest, least am- biguous or confusing means to refer to a character is through evoked information-—information that has had an explicit previous mention. Using evoked information in conjunction with human entities ensures the characters in a story will be clearly recognized and successfully tracked. As for the relatively low use of Evoked sentence topics for episode boundaries, it is difficult to say for sure why this is. Though Evoked sentence topics are used 27% of the time for episode boundaries, it is really evoked abstract entities that occur at a higher percentage at these bound- aries, which suggests that it is more a quality of abstract— ness which is linked to episode boundaries than the level of information used, which is reflected in the statistical analysis presented earlier. 70 4.6.3 Inferred Sentence Topics. Inferred sentence topics occur only 24.1% of the time. But they are heavily correlated with abstract entities (94.7%). They also occur the most frequently at episode boundaries (62.7%), as over half of all inferred sentence boundaries (55.6%) are used as episode boundary markers--all of which are also abstract. Inferred sentence topics also have the lowest topic persis- tence score (0.23). This drops even more with Inferred ab- stract sentence topics (0.03). Based on this information, it seems that Inferred sen- tence topics function mainly as the temporal sequencing in the text. Certainly time is an important element in any narrative, but it is difficult to evoke information about time. Most often, this type of information must be arrived at through inference. Therefore Inferred sentence topics have a primary function of marking episode boundaries. Earlier, statistical evidence was presented that suggested that abstractness is the true trigger for episode boundaries not level of familiarity. However, this overlooks the strong ties between abstract and inferred entities. It may be true that an abstract entity will promote the occurrence of an episode boundary, but it is also true that most fre- quent way to refer to an abstract entity is by an Inferred sentence topic. This will also account for the low topic persistence scores. The temporal framework is something that is only mentioned once, and then is assumed to be presupposed kno tex gre spe. and nar 4.7 use epi. are can thre fun: ter. ter. 1:05 of. 71 knowledge for the text. It would only be in more abstract texts where time would be a topic of discussion for any great length of time--but certainly not in a narrative. The speaker sets the temporal framework and marks a new episode, and then goes on to tell about the important topics in the narrative, the characters. 4.7 Summary A number of tables were presented to illustrate the i use of sentence topics as it correlated with animacy, ) episode boundaries, and topic persistence. Though these are not hard rules to stand by, the following statements can be made which summarize the function of each of the three levels of information. Brand New sentence topics function to begin the text and introduce important charac- ters. Evoked sentence topics function to track the charac- ters through the text. Inferred sentence topics function to set the temporal framework of the text. The above generalizations are based on the analysis of oral texts. The next chapter will analyze written texts. This different medium may cause differences to occur in the way the levels of familiarity are used both in frequency and in correlation with the specified parameters. 5.1 top the tiv: six ten< The top and Per. The the the ter Onlj sen. m; ten( fOUr THE FUNCTION OF SENTENCE TOPICS IN WRITTEN MANDARIN NARRATIVE 5.1 Introduction The preceding chapter discussed the use of sentence topics in oral Mandarin narrative. This chapter will use the same framework but apply it to written Mandarin narra- tive. Therefore, the rest of this chapter will include six sections. The first will present the breakdown of sen- tence topics as they relate to the levels of information. The second will examine how animacy interacts with sentence topics. The next section will discuss episode boundaries and sentence topics. After this section, the role of topic persistence will be applied to the use of sentence topics. The fifth section will, again, draw some conclusions about the fucntion of sentence topics in written narrative. And the last section will summarize the findings of this chap- ter. Throughout the first few sections of this chapter -only superficial comparisons between oral versus written sentence topics will be made; a more complete comparison will be reserved for the sixth section. The individual sen- tence topics for each of the six written texts can also be found with their respective texts in the Appendix. 72 _. 73 5.2 Levels of Familiarity and Sentence Topics in Written Texts As in the previous chapter, before any observations can be made about the use sentence topics in written Man- darin narrative, the patterns of distribution across the levels of familiarity must first be established. Table 10 below, then, is the breakdown of the number of occurrences of a sentence topic per level of familiarity per text. Also, next to each number is the percentage of usage of that particular level of familiarity for each text. Table 10: Written Texts-~Levels of Familiarity BN E I Heshang 0 (0.0%) 26 (63.4%) 15 (36.56) Yanli 1 (5.6%) 9 (50.0%) 8 (44.4%) Duzuo 3 (30.0%) 5 (50.0%) 2 (20.0%) Jixinzi 1 (10.0%) 6 (60.0%) 3 (30.0%) Gechangia 2 (13.3%) 8 (53.5%) 5 (33.3%) ‘Yongshi 1 (7.7%) 9 (69.2%) 3 (23.1%) (Combined 8 (7.5%) 63 (58.9%) 36 (33.6%) The same basic frequency of use of the three levels of familiarity as was noted in the oral texts is the same for tflne written texts. The lowest frequency of use is Brand New Senatence topics (7.5%), followed by Inferred (33.6%). The 1lighest frequency is again Evoked sentence topics (58.9%). Ihlt again, this table really does not say much about why tllese patterns are present, and an examination of the para- nleters of animacy, episode boundaries, and topic persistence iss necessary to provide any insight. 5.3 flu mac" cen 16W tot. tic sak her Gec Tab BN Wit are the Wpi 74 5.3 Animacy and Sentence Topics in Written Texts One of the parameters which seemed to have a major in- fluence on the selection of oral sentence topics was ani- macy. Like Table 2, Table 11 gives the individual per- centage for each level of animacy as it occurs within each level of familiarity. This is based on the total number of occurrences (designated by 'O’ in table 11) divided by the 3 total number of sentence topics which occurred at that par- “ ticular level of familiarity (designated by 'T’). For the i sake of space, each text has been abbreviated as indicated here: Heshang--HE, Yanli--YA, Duzuo—-DU, Jixingzi--JI, Gechangjia--GE, Yongshi--YO. Table 11: Written Texts--Animacy and Sentence Topics HE YA DU JI GE YO O T Percent H 0 1 2 1 2 0 6 8 75.0% 'BN N 0 0 1 0 O 1 2 8 25.0% A 0 0 0 0 0 0 0 8 0.0% H 25 3 5 3 7 9 52 63 82.5% E N 1 5 0 3 0 0 9 63 14.3% A 0 1 0 0 1 0 2 63 3.2% H 0 3 0 0 0 3 36 8.3% I N 1 2 0 1 2 0 6 36 16.7% A 14 3 2 2 3 3 27 36 75.0% 'As with the oral texts, many of the same patterns occur ‘With the written texts. Once more, Evoked sentence topics are strongly linked to human entities, and Inferred sentence t'opics are strongly linked to abstract entities. But there Elire also some differences worth noting. The first is the tile occurrence of Brand New inanimate sentence topics in the written texts, where there were no such occurrences in the 0TB OUT SGT to 75 oral texts. The other difference that needs to be pointed out is the marked increase in the use of Inferred inanimate sentence topics (16.7%) in the written texts, as opposed to their relative infrequency in the oral texts (2.5%). The other way to view this data is to see how the respective levels of animacy pattern within the levels of familiarity. Table 12 provides this information with the number of occurrences of a particular level of animacy noted per level of familiarity, followed by a percentage. Table 12: Written Texts--Animacy and Levels of Familiarity Total BN E I Human 61 6 9.8% 52 85.2% 3 4.9% Inanimate 17 2 10.5% 9 52.9% 6 35.3% Abstract 29 0 0.0% 2 6.9% 27 93.1% Human entities are again strongly tied to Evoked sentence topics, but not as predominantly as in the oral texts as Brand New sentence topics have a greater percentage of use-- 9.8% for written, but only 1.8% for oral. There is also a Inore even balance in the use of inanimate entities. In oral ‘temts, inanimate entities were 94.4% likely to be evoked. Ehat in written texts, though Evoked inanimate sentence to- llics are still the most common means to refer to an inani- nuate entity, their frequency is diminished to 52.9%, while t'llef'r‘equency of use with Inferred sentence topics is in- <3rweased to 35.3%. And again, there is the strong correla- 'ti43n between Inferred sentence topics and abstract entities. di‘ ea:' rez 9P lit HUI le' ba: 001 toi al. Ta‘r Th US$ t9! ta: 18 r 501 fr] th; 76 5.4 Episode Boundaries and Sentence Topics in Written Texts As with the oral texts, each of the written texts were divided into episode boundaries. This proved to be a much easier task than before, since most of the texts were al- ready in paragraphs, which naturally serve to separate one episode from the other. Again, these episode boundaries are listed with their respective texts in the Appendix. Table 13 below gives the text by text breakdown of the number of episode boundaries as they occurred at the three levels of familiarity. On the right are the percentages based on the total number of occurrences of an episode boundary of a particular level of familiarity divided by the total number of occurrences of all episode boundaries (37 in all). Table 13: Written Texts--Episode Boundaries and Levels of Familiarity HE YA DU JI GE YO Total Percent BN 0 1 2 1 2 1 7 18.9% E 7 1 2 1 3 2 16 43.2% I 3 4 1 2 2 2 14 37.8% This table shows that Evoked sentence topics are more often used as episode boundary markers that either Brand New sen- tence topics or Inferred sentence topics.. But again this table should not be interpreted as indicating that only 18.9% of Brand New sentence topics will be used as episode boundaries, but rather that as a whole text, their overall frequency in comparison to other sentence topics is less than 20%. 77 In order to determine the likelihood of any one type 'of sentence topic being used as an episode boundary the total number of occurrences of that level of familiarity at an episode boundary must be divided by the total number of all occurrences of the level of familiarity. This is what Table 14 does. Table 14: Written Texts--Episode Boundaries and Total Occurrences of Levels of Familiarity Episode Boundaries/Occurrences Percentages BN 7/8 87.5% E 16/63 25.4% I 14/36 38.9% Again, like the oral texts, in the written texts a Brand New sentence topic has a high frequency of occurrence at an epi- sode boundary, but at a lower frequency than the oral texts. And there is still the greater chance for an Inferred sen- tence topic to occur at a sentence boundary than a sentence topic which is evoked. But what is different is the more frequent use of Evoked sentence topics at episode boundaries (a fourth of all Evoked sentence topics) in the written texts as opposed to the infrequent occurrence in oral texts: only 6.5%. Further distinctions along these lines can be made by dividing the three levels of familiarity into the three levels of animacy. Table 15 below contains this informa- tion. T: as hu al an en' t8] le‘ te' th Va. So" 9P th. .0" 9D 78 Table 15: Written Texts--Episode Boundaries and Animacy Episode Boundary Total Occurrences Percent H 6 6 . 100.0% BN N 1 2 50.0% A 0 0 0.0% H 13 52 25.0% E N 2 9 22.2% A 1 2 50.0% H 2 3 . 66.7% I N 0 6 0.0% A 12 27 44.4% This table reveals several things. First, it re-emph- asizes the strong correlation of Brand New sentence topics, human or inanimate, at an episode boundary. This table also shows that if an Inferred sentence topic is used at an episode boundary, it will either refer to an abstract entity a human one. This table also shows that Evoked sen- tence topics can be used at episode boundaries with any level of animacy. A final note, is that, unlike the oral texts, inanimate entities can occur at episode boundaries through either Brand New or Evoked sentence topics. ‘As before with the oral texts, a statistical analysis was run on the above data. Because in the written texts, a Brand New sentence topic does not always occur at an epi- sode boundary, and since inanimate entities also occur at episode boundaries, these were included in the analysis for the written texts. The initial analysis produced the fol- lowing probabilities for the above factors to occur at an episode boundary. Ta} Bra Ev< In: Hux Ina Abs The ten so< te: te: t0} Sli re' fa wr fi< is tit th fe‘ 79 Table 16: Written Texts——Probability and Episode Boundaries Probability Brand New 0.895 Evoked 0.224 Inferred 0.289 Human 0.579 Inanimate 0.257 Abstract 0.678 These figures are quite different than the ones for oral. texts. Abstract entities continue to be promoters of epi- sode boundaries, but not as strong as they were in oral texts (.805). Human entities are also promoters in written texts, but not at all in oral (.195). And Inferred sentence topics are no longer promoters at all. More interesting results happen when a stepwise regres- sion is done on the above data. In the oral texts, this regression indicated that level of familiarity was not a factor in the use of episode boundaries. But now, for the written texts, the reverse is true. Animacy is not signi— ficant, but level of familiarity is. What this suggests is that in written texts episode boundaries are not sensi- tive to distinctions in animacy, and that any one of the three types of sentence topics--Brand New, Evoked, or In- ferred-~can be used to signal an episode boundary. Whereas in oral texts, abstract entities were most likely to indi- cate an episode boundary (and in particular Inferred ab- stract entities), in written texts no distinctions can be made--that is to say, no one type of sentence topic is used more than any other to indicate an episode change. There- fore, it can be concluded that oral texts are more rigid 80 in the marking of episode boundaries, while written texts are not. 5.5 Topic Persistence and Sentence Topics in Written Texts The final parameter to be investigated is topic persis- tence. Again, this measures the cataphoric re-occurrence of a topic in successive clauses. The higher the topic persistence, the greater importance that topic plays in the discourse. Table 17 gives the overall topic persistence score of each of the three levels of animacy. This is based on the individual counts at each level for each text. For example, in the Heshang text for Inferred sentence topic. the numbers 2/13 are given. This means that of the thirteen Inferred sentence topics in the Heshang text, they had a to- tal topic persistence score of 2. Table 17: Written Texts--TP and Levels of Familiarity HE YA DU JI GE YO Total BN 0/0 9/1 16/3 7/1 15/2 0/1 47/8 E 112/26 11/9 11/5 26/6 44/8 47/9 251/63 I 2/15 12/8 0/2 7/3 4/5 0/3 25/36 BN TP = 5.88 E TP = 3.98 I TP = 0.69 The most startling aspect of this table is the very high topic persistence of Brand New sentence topics. In fact, Brand New sentence topics have the highest topic per- sistence of the other two levels of familiarity. This is very different from the oral texts, in which the few Brand New sentence topics had much lower topic persistence scores. B.‘ St? Pe te th 0i 81 Even finer distinctions can be determined by once more dividing the three levels of familiarity into the three le- vels of animacy. Table 18 presents this information using the same format as found in Table 17. Table 18: Written Texts——Animacy and Topic Persistence HE YA DU JI GE YO Totals H 0/0 9/1 13/2 7/1 15/2 0/0 44/6 BN N 0/0 0/0 3/1 0/0 0/0 0/1 3/2 A 0/0 0/0 0/0 0/0 0/0 0/0 0/0 H 111/25 6/3 11/5 19/3 43/7 47/9 237/52 E N 1/1 5/5 0/0 7/3 0/0 0/0 13/9 A 0/0 0/1 0/0 0/0 1/1 0/0 1/2 H 0/0 9/3 0/0 0/0 0/0 0/0 9/3 I N 0/1 2/2 0/0 7/1 7/2 0/0 16/6 A 2/14 1/3 0/2 0/2 0/3 0/3 3/27 H TP = 7.33 BN N TP = 1.50 A no occurrence H TP = 4.56 E N TP = 1.44 A TP = 0.50 H TP = 3.00 I N TP = 2.67 A TP = 0.11 The information of Table 18 can be used to group the sentence topic categories into a hierarchy of highest topic persistence to lowest. But unlike the oral topic persis- tence hierarchy, there are a few surprises. Most notably the placement of Inferred sentence topics towards the top of the scale, which will be discussed below. 82 Brand New Human TP = 7.33 Evoked Human TP = 4.56 Inferred Human TP = 3.00 Inferred Inanimate TP = 2.67 Brand New Inanimate TP = 1.50 Evoked Inanimate TP = 1.44 Evoked Abstract TP = 0.50 Inferred Abstract TP = 0.11 Brand New Abstract TP = 0.00 (no occurrence) This breakdown in Table 18 reveals a number of inter- esting facts. The first is that Brand New sentence topics have an even higher topic persistence when a distinction is made between human and inanimate entities. Inanimate enti- ties, used only twice in the texts as Brand New sentence tOpics, have a low topic persistence score. Whereas, human Brand New sentence topics have a very high topic persistence score, which is significantly higher than the next highest sentence topic type, an Evoked human sentence topic. Also of interest is the relatively high topic persis- tence of Inferred inanimate sentence topics. In the oral texts, the topic persistence for this category was 0.00-- one of the lowest scores. But in the written texts, it is 2.67, one of the higher scores. 5.6 Discussion The above tables of information provide a multitude of numbers. And as with the oral texts, it is possible to sift through these numbers and arrive at certain conclusions about the primary function of each of the three levels of information. 83 5.6.1 Brand New Information. In the oral texts, it was noted that Brand New sentence topics are infrequent, and function mainly to get the discourse going. They introduce less important characters, which have a low topic persis- tence—-at least compared to characters which are evoked. Brand New sentence topics in written texts seem to have a much more important role in the discourse. Like in the oral texts, Brand New sentence topics occur dis- course initially at episode boundaries. But in the written texts, Brand New sentence topics seem to function to intro— duce major characters into the discourse, as evidenced by the high topic persistence scores. Like the oral texts, they are most often used with human entities (only twice was a Brand New sentence topic used with an inanimate entity with a subsequently low topic persistence score). Surprisingly, the importance of Brand New sentence topics goes contrary to the statement made in an earlier chapter that a sentence topic must refer to entities, which have a previous mention or which can be deduced through inference. In oral texts, this is not such a problem, as Brand New sentence topics are not really important to the text. They occur infrequently and have fairly low topic persistence scores. Their function, then, seems to be as a way to get the narrative going, but have little other impact on the text as a whole. This is not the case with Brand New sentence topics in written texts. Even though they only occur discourse 84 initially (and so also function as in the oral texts to get the narrative going), Brand New sentence topics also introduce major characters into the text. They occur with greater frequency and have the highest topic persistence score of all sentence topic types. Even though they may have the same function as in the oral texts, Brand New sentence topics appear to have a greater importance. 5.6.2 Evoked Sentence Topics. Evoked sentence topics in oral narrative were said to perform the important func- tion of tracking the characters. -Their high correlation with human entities and their high topic persistence support this claim. And their overall high occurrence in the text indicates how important the characters are in an oral narra- tive. Despite the lower overall frequency of use in the written text (used only 58.9% of the time as compared to 74.7% in oral narratives), Evoked sentence topics still per- form much the same function in the written texts as they do in the oral texts. Like the oral texts, the written Evoked sentence topics are strongly associated with human entities, and have a high topic persistence. Consequently, Evoked sentence topics in written narrative also function to track the characters throughout the discourse. It would also be wrong to conclude that because cer- tain figures are lower in the written text as opposed to the oral text that Evoked sentence topics are not as impor- tant to the written text. Most of the discrepancies between 85 the figures of the written to oral texts can be explained away through the Brand New sentence topics. In written texts, as has been noted above, Brand New sentence topics have a much greater frequency of use, and have very strong correlations with humanness. In a sense, Brand New sentence topics in written texts have taken over or at least have overlapped some of the functions of the Evoked sentence topics. Their greater frequencies and'ties with human entities prevent Evoked sentence topics from being used as much as could be. Evoked sentence topics in written texts are just as important in tracking the characters as they are in oral texts. 5.6.3 Inferred Sentence Topics. In chapter four, the function of Inferred sentence topics was described as a means to mark the temporal episode boundaries. Their low frequency, low topic persistence, and high correlation with_ abstract entities are indicative of this. This is also mostly true for Inferred sentence topics in written texts. There is a high correlation with abstract entities (93.1%), and a low topic persistence (0.69). But this is not the whole story. The telling anomaly is in the greater frequency of use of Inferred sentence to- pics in written texts (33.6%) than in oral texts (24.1%). This can be accounted for by a greater percentage of In- ferred inanimate sentence topics which occur in written texts (5.6%) as opposed to oral texts (0.6%). 86 There is also a much stronger correlation of inanimate entities with Inferred sentence topics in written texts. An inanimate entity is occurs 35.3% of the time with an In- ferred sentence topic--but only 5.6% of the time in an oral text. Neither is it the case that this is due to episode boundaries in written texts, because there are no occur- rences of Inferred inanimate sentence topics at episode boundaries in either oral or written narratives. The key to understanding this may lie in topic persis- tence. Earlier in this chapter, it was remarked that in a ranking of topic persistence from highest to lowest, one of the higher topic persistence scores belongs to Inferred inanimate sentence topics. But in oral texts this category has an extremely low topic persistence score. This suggests that, just as Evoked sentence topics function to track human entities, in written texts Inferred inanimate sentence to- pics function to track inanimate entities--such as the in- scribed board in the Yanli text, or the boiled egg in the Jixingzi text. 5.7 Summary In this chapter the parameters of animacy, episode boundaries, and topic persistence were analyzed to determine how they influenced the use of sentence topics in written narratives. This resulted in several statements about the function of sentence topics as they relate to the levels of familiarity in written texts, and how they were similar or different from the function in oral texts. 87 Brand New sentence topics appear to have a greater im- ‘portance in written texts than in oral texts. They are dis- course initial, but introduce major characters into the nar- rative. Evoked sentence topics were shown to function in the same basic manner in both written and oral texts--that is, they serve to track the characters in the text. Final- ly, Inferred sentence topics were seen to have a much more diverse function in written texts than in oral texts. In both types of text, Inferred sentence topics, especially those that refer to abstract entities, function as episode boundary markers. But also in written texts, Inferred inanimate sentence topics function to track important in- animate objects in the text. CONCLUSION The stated purpose of this paper was to demonstrate that the use of sentence topics in Mandarin oral and written narrative is not arbitrary but determined by certain dis- course and pragmatic parameters. It was also proposed that the different mediums of oral versus written narratives would also be reflected in differences in the way sentence topics function. These two proposals have been substanti- ated. In the last two chapters the examination of animacy, episode boundaries, and topic persistence has provided some insight into what affects the use of sentence topics in oral and written Mandarin narrative. For the most part, and un- derstandably so, both oral and written narratives fall sub- Ject to the same patterns. Thus, both oral and written ‘texts make use of Evoked sentence topics, especially those tJdat refer to human entities, to track the characters of the Imarrative. And both make use of Inferred sentence topics, (BSpecially those that refer to abstract entities, to indi- f the text, though this is a much stronger function in (Dral texts. Since both types of texts are narratives, it Inakes sense that certain patterns will be consistent from one text to the other. 88 89 However, there are other patterns which are more typ- ical for one type of text. Specifically, in written texts there is the increased use of Brand New sentence topics; and there is the use of Inferred inanimate sentence topics to track important inanimate entities in the text. Why are these peculiar to written texts and not to oral texts? The answer would seem to lie in the nature of the text itself. An oral text is an intangible entity in and of it- self. It is non-retrievable, and the narrator must be care- ful not to lose his or her audience with ambiguities and complexities. There is no way for a listener to go back and mull over a difficult passage--the processing of the text must be simple and easy or the story will fail. This, then, almost demands a high use of Evoked sen- tence topics. By topicalizing entities which have had explicit previous mention in the text, the participants are able to quickly discern the referent and have a clear framework in which the remaining sentence will fit into. .And since most stories are about people, not concepts or ‘things, the use of Evoked human sentence topics ensures Tihe clear, understandable flow of information from narrator to listener. Written texts are altogether different. The reader llas a record before him or her of the entire text. The reader has the luxury of returning to earlier passages to reactivate forgotten information about a character or event. He or she can take time to make the correct inferences which 90 the author intended. The written text, then, actually gives the author greater license in how the text is constructed. It is possible to make greater use of Brand New sentence topics to quickly introduce important characters. Certain- ly, the use of a Brand New sentence topic is a quicker way to establiSh an individual framework than having to rely on a previous mention. Though it may result in initial con- fusion, the reader can always return to this passage to gain any clarity that may be lacking. Since the written text is retrievable, it is not as important to make constant mention of the characters through sentence topics. Several clauses can be devoted to an important inanimate object without too much fear that the readers will lose track of the characters. And it is alSO possible to establish this referent through in- ference much more easily than in an oral text. An infer- ence relies on the logical or common sense processing of the participant. But because the information flows so quickly in an oral narrative, he or she may not have the necessary time to establish the referent before additional information is being added. This is not a problem in a written text, where the reader can take as long as need be to clearly understand whatever is being inferred. This is not to say that the author will always use an Inferred sentence topic, only that he or she has option to do so much more than the narrator of an oral text. 91 Clearly, in these texts there is something which is triggering an increase usage of Brand New and Inferred sen- tence topics in written texts. Obviously, much the above account for this is speculation, and it is not possible to say with complete surety why these patterns exist. It has 'been possible, however, by using quantitative discourse .analysis to gain a better understanding of the functions of sentence topics as used in narratives. Neither is it an arbitrary process, but one which is dependent upon and is influenced by various discourse parameters, such as animacy, episode boundaries, and topic persistence. Consequently, this analysis gives support to Chafe’s notion of a sentence topic as setting a temporal, location- .al, or individual framework for the sentence. At least for (Jhinese, sentence topics, which have been chosen according tn: Chafe’s model, have been shown to be sensitive to various ciiscourse parameters. If this was not the case, then it (Mould be said that Chafe’s conception of sentence topic is rust a true reflection of language. But the evidence shows tJnat these 'Chafean’ sentence topics do indeed function in 1language. And these are distinctions that other topical Studies in languages like Japanese and Chamarro (as reported irl Myhill 1992: ch. 2, pp. 23-26) have not been able to make ‘yven.though they too made use of parameters like animacy and tOpic persistence . APPENDIX THE TEXTS In the Appendix, the three oral and six written texts are given in their entirety. Preceeding each narrative is a brief summary of the plot. This is followed by the actual text with an English word for word gloss and a free transla- tion for each line. A line, which represents an episode boundary, will be indentified with the notation EB following the line number. After the end of each line the sentence topic will be identified as well as the corresponding level of familiarity, animacy, and topic persistence score. For the sake of brevity and simplicity, only the pertinent in- formation for each sentence topic will be given by means of a specified notation. The following is a list of those no- tations, and what each refers to. EN: Brand New I ( )/ : Inferrable (entity inferrable fromtype)/inference type IC ( )/ : Containing inferrable (contained entity, inferrable fromtype)/infer- ence type ' E: Evoked (textually) SA: Stereotypic assumption For example, in line 10 of the Dushuren text, the sentence topic is encoded by the noun phrase liang ge xiao nuer ‘the 92 93 two smallest daughters.’ The notation for this sentence topic would be: I (si ge nuerE)/part-whole. This means that this sentence topic is an inferrable. Furthermore, the entity from which it is inferred is encoded by the ’ which represents noun phrase Si ge nuer 'four daughters, an evoked entity (from line 1 of the text). Lastly, the inference is a part to whole relationship. The Da Hui Lang Text This is the longest of the three oral narratives, run- ning over a hundred lines. The story concerns two little girls who were abandoned by their father in the mountains. They chance upon a hollow tree in which an old grandmother is living. She tells the little girls that a wolf also lives in that place, but agrees to hide them. The wolf comes home and after awhile realizes that there are more people in the tree then there should be. He attempts to find the two little girls, but fails. After he falls asleep, the old grandmother and the two little girls kill the wolf. And they all live happily ever after. Da Hui Lang 001 (EB) You yi jiaren, ta jia li you si ge nuer. have one family 38 family in have four CLSF daughter There was a family which had four daughters. topic: you yi jiareni 'one family’ entity: BN Human, TP = 4’ Da Hui Lang 002 You yi tian ta mama dui ta baba shuo, "Women have one day 33 mother to 3s father say 1pl jia you si ge nuer, women mei you name family have four CLSF daughter 1pl not have like that duo dongxi, yang tamen. many thing feed 3pl One day the mother said to the father, "We have four daughters, but we don’t have much food to feed them. topic: you yi tian ’one day’ entity: I (jiareni)/SA: characters must function in time Abstract, TP = 0 94 1,51. ’1 "H 95 Da Hui Lang 002 (con’t) topic: women jiai 'our family’ entity: E Human, TP = 2 Da Hui Lang 003 Women yao ba liang ge nuer gei renjia." 1pl want DO two CLSF daughter give other people We should give two daughters to someone." topic: womeni 'we’ entity: E Human, TP = 0 Da Hui Lang 004 Ta baba shuo, "Shei yao nuer, ruguo shi erzi jiu 3s father say who want daughter if is son then hao le." good CRS The father said, "Who wants a daughter, if it was a son, that would be good." topic: ta babaj 'father’ entity: E Human, TP = 5 topic: shei who entity: E Human, TP II C Da Hui Lang 005 (EB) Suoyi zai zhe zhong qingkuang xia ne, ta baba therefore at this heavy situation under PRTL 3s father jueding ba liang ge nuer dui qu kan chai. decide DO two CLSF daughter take go out firewood Because of this difficult situation, the father decided to take two daughters with him to cut some firewood. topic: zai zhe zhong qingkuang xia ne 'under this diffi- cult situation’ entity: I (line 5)/proposition=situation Abstract, TP = 0 96 Da Hui Lang 006 Ranhou ba tamen reng zai shan shang. then DO 3pl leave at mountains on Then he will leave them in the mountains. topic: ranhou 'then’ entity: I (line 5)/SA: progression of time Abstract, TP = 0 ' Da Hui Lang 007 (EB) Dier tian yi da zao ne, ta baba jiu dui second day one early morning PRTL 3s father then to suoyou de neur shuo, "Baba jintian yao kan all GEN daughter say father today want cut chai. firewood The next day, early in the morning, the father said to all of his daughters, "Today, father will be cutting some firewood. topic: dier tian yi da zao ne 'early in the morning of the second day’ entity: I (preceeding lines)/SA: progression of time Abstract, TP = 1 topic: babaj 'father’ entity: E Human, TP = 5 Da Hui Lang 008 Shan shang you hen duo xian hua. mountains on have very many fresh flower On the mountains there are many wild flowers. topic: shan shang 'on the mountain’ entity: E Inanimate, TP = 5 Da Hui Lang 009 7 Shei yuanyi qu cai hua jiu he baba yiqi who willing go pick flower then with father together qu u go Whoever wants to pick flowers can come with father." 97 Da Hui Lang 009 (con’t) topic: shei 'who’ entity: E ' Human, TP = 8 Da Hui Lang 010 Liang ge xiao nuer bu dongshi. two CLSF small daughter no understand The two smallest daughters did not understand. topic: liang ge xaio nuerk ‘the two smallest girls’ entity: _I (si ge nuerE)/part to whole Human, TP = 5 Da Hui Lang 011 Tamen feichang xihuan hua, jiu shuo, "Baba, baba, 3pl very much like flower then say father father wo yao qu cai hua." 1s want go pick flower They like flowers very much and said, "Father, father, want to pick flowers." topic: tamenk 'they’ entity: E Human, TP = 6 topic: wok 'I’ entity: E Human, TP =5 Da Hui Lang 012 Disan ge nuer he disi ge nuer he ta third CLSF daughter with fourth CLSF daughter with 35 baba yiqi shang shan qu kan chai. father together on mountains go cut firewood The third and fourth daughters went with their father into the mountains to cut firewood. topic: . disan ge nuer he disi ge nuer he ta baba yiqi - k theirJfather’ entity: E Human, TP = 6 'the third and fourth daughter and 98 Da Hui Lang 013 (EB) Zai shan shang de shihou ne, ta baba shuo, at mountains on RLV time PRTL 3s father say "Nimen liang ge zai zhe ge shan shang kan 2pl two CLSF at this CLSF mountains on cut cai. firewood When they got to the mountains, the father said, "You two cut firewood on this mountain. topic: zai shan shang de shihou ne 'when they arrived on the mountain’ entity: IC (shihou, shan shang)/SA: progression of time Abstract, TP = 0 topic: nimen liang gek 'you two girls’ entity: E Human, TP = 5 Da Hui Lang 014 Baba dao duimian de shan shang qu kan chai, father to opposite GEN mountains on go out firewood yinwei duimian shan shang de chai bijiao because opposite mountains on RLV firewood relatively duo. many Father will go to the other side of the mountain to cut firewood, because there is relatively more firewood there. topic: babaj 'father’ entity: E Human, TP = 0 topic: chai tthe firewood’ entity: E Human, TP = 1 Da Hui Lang 015 Zher ne xian hua bijiao duo, nimen keyi zai here PRTL fresh flower relatively many 2pl can at zher cai hua." here pick flower Here are many wildflowers for you to pick." 99 Da Hui Lang 015 (con’t) topic: zher 'here’ entity: E Inanimate, TP = 1 Da Hui Lang 016 (EB) Liang ge nuer ne xinggao-cailie de zai zhai two CLSF daughter PRTL jubilant MNR DUR pick hua. flower The two daughters were excitedly picking flowers. topic: liang ge nuer nek ‘the two daughters’ entity: E Human, TP = 5 Da Hui Lang 017 Tamen yizhi zhai yizhi zhai. 3pl all the time pick all the time pick They were picking flowers constantly. topic: tamenk 'they’ entity: E Human, TP = 4 Da Hui Lang 018 Zhongwu guo qu le. noon pass go PRF Noon passed. topic: Zhongwu ’noon’ entity: I (preceeding lines)/SA: progression of time Abstract, TP = 0 Da Hui Lang 019 Xiawu lai le. afternoon come PRF Afternoon came. topic: xiaowu 'afternoon’ entity: I (line 18)/SA: progression of time Abstract, TP = 1 100 Da Hui Lang 020 Xiaowu ye 'guo qu le. afternoon also pass go PRF The afternoon also passed. topic: xiaowu 'afternoon’ entity: E Abstract, TP = 0 Da Hui Lang 021 Huanghu lai le. . To dusk come PRF i Dusk came. ' 1 topic: huanghu 'dusk’ ‘ entity: I (line 20)/SA: progression of time Abstract, TP = 0 Da Hui Lang 022 Tian jianjian de heixiao lai le. sky gradual MNR darken come PRF The sky gradually darkened. topic: tian 'sky’ entity: I (huanghuI)/SA: dusk entails a darkening sky Abstract, TP = 1 Da Hui Lang 023 Keshi tamen zhe hui you er you ke, but 3pl this time have hungry have thirsty yizhi zai deng zhe tamen de baba lai jie all the time DUR wait DUR 3pl GEN father come pick.up tamen hui jia. 3pl return family But by this time they were hungry and thirsty and waiting for their father to pick them up and to home. topic: tamenk 'they’ entity: E Human, TP = 6 101 Da Hui Lang 024 Keshi zenme deng ye deng bu dao ta baba. but how long wait also wait no until 33 father But no matter how long they waited, their father did not come. topic: ¢k entity: E Human, TP = 4 Da Hui Lang 025 Yizhi dao hen wan, tian dou chanbu heile de all the time until very late sky all whole dark RLV shihou, ta baba hai mei you lai. time 33 father still not have come Meanwhile it was getting late and the sky dark, and their father had still not come. topic: yizhi 'all the time’ entity: I (line 24)/SA: progression of time Abstract, TP = 0 Da Hui Lang 026 (EB) Zhe shihou ne, liang ge xiao nuhai feichang de this time PRTL two CLSF small girl very much MNR haipa, yinwei shan shang you hen duo lang, ye scare because mountains on have very many wolf also you laohu. have tiger By this time, the two little girls were very frightened, because in the mountains there are many wolves and tigers. topic: zhe shihou ne 'this time’ entity: I (preceeding line)/SA: progression of time Abstract, TP = 0 topic: shan shang 'on the mountain’ entity: E . Inanimate, TP = 2 Da Hui Lang 027 102 Tamen haipa lang he laohu lai chi tamen, erque 3pl scare wolf and tiger come eat 3pl in.addition tamen bu zhidao zenme hui jia. 3pl no know how return home They were scared that the wolves and tigers would eat them, additionally, they did not know how to get home. topic: tamenk 'they’ entity: E Human, TP = 6 topic: tamenk 'they’ entity: E Human, TP = 5 Da Hui Lang 028 Tamen jiejie 3pl older sister to haipa; women shun zhe, . dui meimei shuo, "Bu yao younger sister say no want zhe ge shanlu wang scare 1pl along this this CLSF mountain path toward xiao zou, yiding hui zhaodao jia de." foothill walk certainly able find home PRTL The older sister said to the younger one, "Don’t be scared, we will follow this path to the foothills, and we will certainly be able to find our way home." topic: tamen jiejiel entity: E Human, TP = 7 topic: )6“) E 'the older sister’ entity: Human, TP = 9 topic: womenk 'we’ entity: E Human, TP = 4 Da Hui Lang 029 (EB) Zhe shihou, zhe ge jiejie ne' dai zhe this time this CLSF older sister PRTL take DUR meimei yi lu zou yi lu xiang zhaodao younger sister one road walk one road think find 103 Da Hui Lang 029 (con’t) tamen de jia, keshi zenme ye zhao bu dao tamen de 3pl GEN home but what also look no arrive 3pl GEN jia. home At this time, the older sister led the younger sister down the path, thinking they would find their home, but what they looked for, they did not find. topic: zhe ge jiejiel ne 'this older sister’ entity: E Human, TP = 6 topic: ¢k E entity: Human, TP = 2 Da Hui Lang 030 Tamen yizhi zou yi zou yi, zou dao 3pl all the time walk one leave one walk to ban-shanyao de yi ke da shu dixia. half way up a mountain RLV one CLSF big tree on ground While they were walking, they came to the bottom of a big tree that was half way up the side of the mountain. topic: tamenk 'they’ entity: E Human, TP = 1 Da Hui Lang 031 Meimei shizai lei 1e, zuo xiao 1e dui younger sister really tired CRS sit down PRF to jiejie shuo, "Jiejie wo hen lei, wo zai older sister say older sister ls very tired 1s again ye zou bu dong le." also walk no move CRS The younger sister is very tired, she sat down and said to the older sister, "Older sister, I am very tired, I cannot walk any farther." topic: meimei 'younger sister’ entity: E Human, TP = 9 104 Da Hui Lang 031 (con’t) topic: wom 'I" entity: E Human, TP = 8 Da Hui Lang 032 Jiejie shuo, "Ni yiding de qilai zou ya, ni older sister say 23 certainly MNR go up walk PRTL 23 bu zou de hua ne, laohu hui lai chi ne." no walk PRTL if PRTL tiger will come eat PRTL The older sister said, "You must get up, if you don’t, a tiger will come and eat you." topic: jiejiel 'older sister’ entity: E Human, TP = 5 topic: him 'you’ entity: E Human, TP‘= 8 Da Hui Lang 033 Suoyi meimei mei you banfa, yizhi therefore younger sister not have means all the time he jiejie zou. with older sister walk Therefore the younger sister continued to walk with the older sister. topic: meimei 'younger sister’ entity: E Human, TP = 6 Da Hui Lang 034 (EB) Zou dao qianbian de shihou ne, turan jian zai walk to in front of RLV time PRTL suddenly meet at yi ke shu dixia kanjian you yidian one CLSF tree on ground catch sight of have little liangguang. light After walking awhile, they suddenly came to a tree from which they noticed a small light was coming out of. 105 Da Hui Lang 034 (con’t) topic: zou dao qianbian de shihou ne 'after they had walked awhile’ entity: IC (shihou, line 33)/SA: progression of time Abstract, TP = 0 Da Hui Lang 035 Jiejie shuo, "Ya, zhe shi yi jia renjia. older sister say PRTL this is one CLSF household The older sister said, "This is a house. topic: jiejiel 'older sister’ entity: E Human, TP = 5 topic: zhe 'this’ entity: E Inanimate, TP = 3 Da Hui Lang 036 Women jinqu kan yi kan. 1pl enter see one see Let’s enter and look around a little. topic: womenk ’we entity: E Human, TP = 3 Da Hui Lang 037 Women you jiu le." 1pl have save CRS We have been saved." topic: womenk 'we’ entity: E Human, TP = 2 Da Hui Lang 038 Zhe shihou, tamen faxian zai shu gen dixia you this time 3pl find at tree with on ground have yi ge dong. one CLSF cave At this time, they found a cave at the bottom of the tree. 106 Da.11ui Lang 038 (con ’t) txapic: zhe shihou 'this time’ «entity: I (preceeding line)/SA: progression of time Abstract, TP = 0 Da Hui Lang 039 INa ge dong limian fachu yidian weiruo de dengguang. that CLSF cave inside shine little soft ADV lamplight Inside the cave a small light softly shone. topic: na ge dong 'that cave’ entity: E Inanimate, TP = 2 13a Hui Lang 040 Jiejie ba dong de gaizi jiekai, he older sister DO cave GEN cover open with meimei yiqi jinqu. younger sister together enter The older sister opened the door to the cave, and with the younger sister entered. topic: jiejiel 'older sister’ entity: E Human, TP = 3 13a Hui Lang 041 (EB) Zhe shihou ne, zai zhe ge dong libian you yi ge this time PRTL at this CLSF cave inside have one CLSF hen lao hen lao de lao nainai. very old very old RLV old grandmother At this time, inside the cave was a very old grandmother. topic: zhe shihou ne 'this time’ entity: I (preceeding line)/SA: progression of time Abstract, TP = 0 Da Hui Lang 042 Lao nainai toufu hen chang, chuan tou dou shi old grandmother hair very long all head all is baifa, ye mei you shutou, yifu ne feichang de white also not have comb cloth PRTL very.much PRTL 107 Da Hui Lang 042 (con’t) polan. shabby The old grandmother had hair that was very long, white, and uncombed, and her clothes were very shabby. topic: lao nainai 'old grandmother’ entity: E Human, TP = 7 n Da Hui Lang 043 Ta zuo zai hunan de deng xia ne fengbu yichang. 3s sit at dim RLV light under PRTL sewing cloth She was sitting under a dim light sewing. topic: t she entity: E Human, TP = 4 I)a Hui Lang 044 Ta kanjian liang ge xiao nuhai jinlai yihou 3s catch sight of two CLSF small girl enter after wen, "Shui jiao nimen jinlai de ya?" ask who ask 2pl enter PRTL PRTL She saw the two little girls enter and asked them, "Who asked you to enter?" topic: ta kanjian liang ge xiao jinlai yihou 'after she saw the two little girls enter’ entity: I (line 43)/SA: progression of time Abstract, TP = 0 topic: shui who entity: I (unspecified character)/SA: someone must give permission to enter Human, TP = 0 Da Hui Lang 045 Xiao nuhai shuo, "Lao nainai jiujiu women. small girl say old grandmother save 1pl The little girls said, "Old grandmother save us. topic: xiao nuhaik 'little girls’ entity: E Human, TP = 7 topic: 1ao nainai 'old grandmother’ entity: E Human, TP = 3 n 108 Da Hui Lang 046 Women mi lu le, bu zhi zenme hui jia." 1pl lose road CRS no know how' return home We have lost our way and don’t know how to return home." topic: womenk 'we’ entity: E Human, TP = 7 Da Hui Lang 047 Lao nainai shuo, "Zhe bu shi ren zhu de old grandmother say this no is person live RLV difang. place The old grandmother said, "No person lives here. topic: lao nainai 'old grandmother’ entity: E Human, TP = 3 11 topic: zhe 'this’ entity: E Inanimate, TP = 1 Da Hui Lang 048 Zhe shi da hui lang zhu de difang." this is big grey wolf live RLV place A big grey wolf lives here." topic: zhe 'this’ entity: E- - Inanimate, TP = 0 Da Hui Lang 049 Liang ge xiao nuhai xia de bu de le, shuo, two CLSF small girl scare PRTL fear PRTL CRS say "Lao nainai women mei you bunfa, women bu renshi old grandmother 1pl not have way 1pl no know lu hui jia. road return home The two little girls were scared to death and said, "Old grandmother, we do not know how to get home. 109 Da Hui Lang 049 (con’t) topic: entity: E Human, topic: womenk entity: E Human, Da Hui Lang 050 Ni neng bu neng 23 can no can liang ge xiao nuhaik 'the two little girls’ TP = 6 (we: TP = 6 yi wan? live one night rang women zhu allow 1pl Can you let us stay the night? 1 topic: nin you entity: E Human, TP = 4 Da Hui Lang 051 Women mingtian zai tomorrow again return home 1pl hui jia." Tomorrow, we will go home." topic: womenk entity: E Human, Da Hui Lang 052 Lao nainai old grandmother say lang yihui jiu wolf soon (we, TP = 5 "Zhe ke this can no bu xing ya, da hui do PRTL big grey shuo, huilai le. immediately return CRS The old grandmother said, "You cannot do this, the big grey wolf will soon return. topic: lao nainai 'old grandmother’ n entity: E Human, TP = 3 topic: zhe 'this’ entity: I (lines 49-51)/proposition=situation Abstract, TP = 0 110 Da Hui Lang 053 Ta wen dao ren wei, hui yao lai chi nimen de." 33 smell to person smell will want come eat 2pl PRTL He will smell you and want to eat you." topic: tao 'he’ entity: E Human, TP = 2 Da Hui Lang 054 Liang ge xiao nuhai xia de hun sheng fudou. two CLSF small girl scare MNR whole body shivering The two little girls shivered with fright. topic: liang ge xiao nuhaik 'the two little girls’ entity: E . Human, TP = 8 Da Hui Lang 055 (EB) Zhe shihou ne, zhe ge lao nainai this time PRTL this CLSF old grandmother kanjian zhe liang ge haizi shizai kelian, jiu catch sight of this two CLSF child really pity then dui zhe liang ge haizi shuo, "Bu yong baipa, wo ba to this two CLSF child say no use afraid 13 DO nimen cangqilai. 2pl hide Now the old grandmother saw these two children and had pity, and said to them, "Do not worry, I will hide you. topic: zhe shihou as ’this time’ entity: I (preceeding propostion)/SA: progression in time Abstract, TP = 0 topic: entity: E Human, TP = 8 Da Hui Lang 056 Da hui lang lai de shihou, nimen jiu duo zai big grey wolf come RLV time 2pl. then hide at nimen de difang bu yao zuo sheng." 2pl GEN place no want make noise 111 Da Hui Lang 056 (con’t) When the big grey wolf comes, you hide in your place and not make any noise." topic: da hui lang lai de shihou ne 'when the big grey wolf comes’ ' entity: I (line 52)/SA: the wolf will return at some time Abstract, TP = 0 Da Hui Lang 057 Lao nainai nachu 1e xian guo he shiwu rang old grandmother bring PRF fresh fruit and food. allow zhe liang ge xiao nuhai chi. this two CLSF small girl eat The old grandmother brought them fresh fruit and food and let the two little girls eat. topic: lao nainai 'old grandmother’ entity: E Human, TP = 4 n Da Hui Lang 058 Zhe liang ge xiao nuhai chi le hao duo hao duo this two CLSF small girl eat PRF well many well many de dongxi. MNR thing The two little girls eat a lot of food. topic: zhe liang ge xiao nuhaik 'the two little girls’ entity: E Human, TP = 7 Da Hui Lang 059 (EB) Zhe hui ne, tamen turang tingdao waimian you this time PRTL 3pl suddenly hear outside have jiaobu sheng. foot noise Then they suddenly heard footsteps outside. topic: zhe hui ne 'this time’ entity: I (preceeding propositions)/SA: progression of time Abstract, TP = 0 112 Da Hui Lang 060 Lao nainai shuo, "Kuai lai, kuai lai, duo zai yi old grandmother say hurry come hurry come hide at one ge da guizi limian. CLSF big cabinet inside The old grandmother said, "Hurry up, hurry up, hide inside the big cabinet. topic: lao nainai 'old grandmother’ entity: E Human, TP = 1 n topic: 0? entity: E ll 0') Human, TP Da Hui Lang 061 Zhe ge da guizi limian ya, zhe ge da hui this CLSF big cabinet inside PRTL this CLSF big grey lang bu hui qu kai guizi." wolf no will go open cabinet The big grey wolf will not open the cabinet." topic: zhe ge da guizi limian ya ’inside this big cabinet’ entity: E Inanimate, TP = 2 Da Hui Lang 062 (EB) Ranghou ne, zhe ge lao nainai gei zhe liang then PRTL this CLSF old grandmother give this two ge xiao nuhai yi ren yi ge zhuizi. CLSF small girl one person one CLSF needle Then the old grandmother gave each little girl a needle. topic: ranghou ne 'then’ entity: I (line 61)/SA: progression of time Abstract, TP = 0 IDa Hui Lang 063 (EB) Zhe liang ge haizi na zhe zhuizi yihou ne jiu this two CLSF child take this needle after PRTL then duo zai da guizi de yifu houmian. hide at big cabinet GEN cloth behind 113 Da Hui Lang 063 (con’t) After the two children took the needles, they hid behind the clothes cabinet. topic: zhe liang ge haizi na zhe zhuizhi yihou ne 'after the two little girls took the needles’ entity: I (line 62)/SA: progression of time Abstract, TP = 0 Da Hui Lang 064 Zhe hui, da hui lang, "dong, dong" de jinlai le. this time big grey wolf dong dong PRTL enter PRF Then the big grey wolf said "ding, dong" and entered. topic: zhe shihou 'this time’ entity: I (preceeding line)/SA: progression of time Abstract, TP = 0 Da Hui Lang 065 Jiu shuo, "Ha, ha, wo hen er, wo hen er." then say ha ha 13 very hungry 13 very hungry Then he said, "Ha, ha, I am very hungry." topic: go entity: Human, TP = 9 topic: woo entity: E Human, TP = 8 Da Hui Lang 066 Lao nainai shuo, "Ni zuo zai na, wo gei ni old grandmother say 23 sit at there 13 give 23 wanfan." supper The old grandmother said, "Sit down there and I will get you supper." topic: lao nainai 'old grandmother’ _ n entity: E Human, TP = 3 topic: nio ‘you’ entity: E Human, TP = 8 114 Da Hui Lang 067 Da hui lang zuo xialai. big grey wolf sit down The big grey wolf sat down. topic: da hui lango ‘big grey wolf’ entityz. E Human, TP = 6 Da Hui Lang 068 Lao nainai gei ta duan 1e yi bei cha he hen old grandmother for 33 bring PRF one CLSF tea and very duo rou. many meat The old grandmother brought him a cup of tea and a lot of meat. topic: 1ao nainain 'old grandmother’ entity: E Human, TP = 3 Da Hui Lang 069 (BB) Da hui lang chi le rou yihou ne, jiu jue de zhe big grey wolf eat PRF meat after PRTL then feel MNR this ge jia li you yi gu ren wei. CLSF home in have one CLSF person smell After the big grey wolf ate the meat, he sensed that in the house was the scent of a person. topic: da hui lang chi 1e rou yihou ne 'after the big grey wolf ate the meat’ entity: I (line 68)/SA: progression of time Abstract, TP = 0 Da Hui Lang 070 Ta daochu wen ya wen ya xiang zhaodao ren. 3s everywhere smell PRTL smell PRTL want find person He smelled everywhere trying to find the person. topic: tao ’he’ entity: E Human, TP = 4 115 Da Hui Lang 071 Ta wen lao nainai, "Zhe jia li hen qiquai 3s ask old grandmother this home in very strange haoxiang you ren lai seems guo." have person come EXP He asked the old grandmother, "There is something strange in this house-—it seems there is somebody here.' topic: tao ‘he’ entity: E Human, TP = 4 topic: zhe jia 11 entity: E Inanimate, TP = 4 13a Hui Lang 072 Lao nainai shuo, old grandmother say how jiu shi ni he wo ma? just is 23 and 13 INTR The old grandmother said, here, topic: lao nainai n entity: E Human, TP = 1 topic: ¢ entity: E Inanimate, TP = 2 Da Hui Lang 073 Hai you beiren ma?" also have others INTR Who could it be?" topic: ¢ entity: E Inanimate, TP = 1 Da Hui Lang 074 Zhe ge da hui this CLSF big grey wolf say The big grey wolf said, lang shuo, "Zenme hui able have person come all "Bu dui, no 'in this house’ you ren lai, 'old grandmother’ bu dui. right no right "That’s not right. dou "How can there be somebody all there is just you and me? 116 Da Hui Lang 074 (con’t) topic: zhe ge da hui lango 'the big grey wolf’ entity: E Human, TP = 4 Da Hui Lang 075 Zhe limian you ren wei." this inside have person smell There is the scent of a person in this house." topic: zhe limian 'the inside’ entity: E Inanimate, TP = 2 Da Hui Lang 076 Ta daochu zhao. 3s everywhere look He looked everywhere. topic: tao 'he’ entity: E . Human, TP = 5 Da Hui Lang 077 Zhe shihou ta ba dugui dakai. this time 33 DO cabinet open Then, he opened the cabinet topic: zhe shihou 'this time’ entity: I (preceeding propositions)/SA: progression of time Abstract, TP = 0 Da Hui Lang 078 Ta wen, "Zhe limian haoxing you ren wei ya." 33 ask this inside seem have person smell PRTL He said, "This seems to have a person’s scent inside of it." topic: tao 'he’ entity: E Human, TP = 4 topic: zhe limian 'the inside’ entity: I Inanimate, TP = 1 __L .‘s'fl 1" . I a.“ '5 117 Da Hui Lang 079 _ Lao nainai shuo, "Mei you de, mei you de." old grandmother say not have PRTL not have PRTL The old grandmother said, "No it doesn’t." topic: lao nainai 'old grandmother’ entity: E Human, TP = 0 n Da Hui Lang 080 Dangshi, yinwei fangzi li bijiao hei, ta kan bu then because house in relatively dark 33 see no jian guizi li shifou duo 1e ren. see cabinet in if hide PRF person Because it was relatively dark in the house, he was not able to see if a person was hiding inside the cabinet. topic: dangshi 'then’ entity: I (line 79)/SA: progression of time Abstract, TP = 0 topic: tao 'he’ entity: E Human, TP =6 Da Hui Lang 081 Ta jiu yong shou qu mo. 33 then use hand go feel He then felt around with his hand. topic: tao 'he’ entity: E Human, TP = 5 Da Hui Lang 082 A Xiao meimei yong zhuizi zha le yi xiao da small younger sister use needle punch PRF one CLSF big hui lang de shou. grey wolf GEN hand The younger sister poked the big grey wolf in the hand with the needle. topic: xiao meimei ‘the little sister’ entity: E Human, TP = 3 118 Da Hui Lang 083 Da hui lang shuo, "Ai ya, hao tong ya. big grey wolf say ai ya very painful PRTL The big grey wolf said, "Ouch, that’s painful. topic: da hui lango 'the big grey wolf’ entity: E Human, TP = 4 topic: entity: I (line 81)/proposition=situation Abstract, TP = 0 Da Hui Lang 084 Yiding shi dingzi zha le wo de shou." certainly is nail punch PRF 13 GEN hand A nail must have poked my hand." topic: entity: E Inanimate, TP = 0 Da Hui Lang 085 Suoyi, ta gankui ba gui men guanshang, jiu qu therefore 33 hurry DO cabinet gate shut then go shuijiao qu le. sleep go PRF Therefore, he quickly shut the door of the cabinet, and then went to sleep. - topic: tao 'he’ entity: E Human, TP = 5 Da Hui Lang 086 (EB) Dang da hui lang shui de huhu de shihou, lao when big grey wolf sleep MNR good sleep RLV time old nainai ba guizi dakai shuo, "Kuai chu lai, grandmother DO cabinet open say hurry set out come kuai chu lai. hurry set out come When the big grey wolf was sound asleep, the old grandmother opened the cabinet door and said, "Come out, come out. 119 13a Hui Lang 086 (con’t) topic: dang da hui lang shui de huhu de shihou ‘when the big grey wolf was sound asleep’ entity: IC (shihou, da hui lang shui)/SA: progression of time Abstract, TP = 0 tOpic: ¢k entity: E Human, TP = 6 I)a.Hui Lang 087 Women xiang ge banfa ba zhe ge da hui lang 1pl think CLSF means DO this CLSF big grey wolf shadiao." kill We must think of a way to kill the big grey wolf." (we, topic: womenk’n entity: E Human, TP = 3 D8. Hui Lang 088 Suoyi tamen zhao le yi ge shensuo. therefore 3pl look PRF one CLSF rope Therefore, they looked for a rope. topic: tamenk’n 'they’ entity: E Human, TP = 2 Da Hui Lang 089 Ba zhe ge da hui lang nesi le. DO this CLSF big grey wolf hang PRF They hung the big grey wolf. topic: ¢k,n entity: E Human, TP = 1 Da. Hui Lang 090 (EB) Tamen sha 1e da hui lang yihou, da hui lang jia 3pl kill PRF big grey wolf after big grey wolf home li you hen duo shiwu, you haizao, you xian guo, in have very many food have dates have fresh fruit 120 IDa Hui Lang 090 (con’t) hai you qita de shuiguo, gan shuiguo. also have other PRTL fruit dry fruit After they killed the big grey wolf, there was a lot of food in his house--dates, fresh fruit, and dried fruit. topic: tamen sha 1e da hui lang yihou 'after they killed the big grey wolf’ entity: I (line 89)/SA: progression of time Abstract, TP = 0 I)a.Hui Lang 091 Suoyi, zhe liang ge xiao nuer ye bu chou therefore this two CLSF small daughter also no worry chi, bu chou chuang. eat no worry cloth Therefore, the two little daughters did not worry about what to eat or what to wear. topic: zhe liang ge xiao nuerk 'the two little daughters’ entity: E Human, TP = 6 I)a Hui Lang 092 Lao nainai shuo, "Nimen liang ge keyi gen wo old grandmother say 2pl two CLSF can with 13 zhu zai yiqi. live at together The old grandmother said, "You can live with me. topic: lao nainai 'old grandmother’ entity: E Human, TP = 5 n topic: nimen liang gek 'you, two’ entity: E ‘ Human, TP = 8 I)a.Hui Lang 093 Wo xiang nimen de baba, mama shi bu xiangyao nimen 13 think 2pl GEN father mother is no want 2pl le." CRS I don’t think your mother and father want you." 121 Da Hui Lang 093 (con’t) topic: won 'I’ entity: E Human, TP = 3 jDa Hui Lang 094 Suoyi, zhe liang ge xiao nuhai jiu he zhe therefore this two CLSF small girl then with this ge lao nainai zhu zai yiqi. CLSF old grandmother live at together Therefore, the two little girls then stayed with the old grandmother. topic: zhe liang ge xiao nuhaik 'the two little girls’ entity: E Human, TP = 8 Da Hui Lang 095 (EB) Liang nian guoqu le. two year pass PRF Two years passed. topic: 1iang nian 'two years’ entity: I (line 93)/SA: progression of time Abstract, TP = 0 IDa.Hui Lang 096 Zhe liang ge nuhai zhang de da le yidian. this two CLSF girl grow MNR big PRF little The two girls had grown a little bigger. topic: zhe liang ge nuhaik ’the two girls’ entity: E Human, TP = 9 Da Hui Lang 097 - Tamen shuo, "Women yao hui jia kan women de baba, 3pl say 1pl want return home see 1pl GEN father mama." mother They said, "We want to go home and see our mother and father." .__.___..» 1.011: _r it . I p. 122 Da Hui Lang 097 (con’t) topic: tamenk 'they’ entity: E Human, TP = 8 topic: womenk entity: E Human, TP = 7 LDa Hui Lang 098 Lao nainai shuo, "Wo keyi dai nimen xiao shan, old grandmother say 13 can take 2pl down mountains danshi, nimen yiding de hui lai, yinwei nimen but 2pl certainly ADV return come because 2pl de baba, mama yijing bu ai nimen le." GEN father mother already no like 2pl CBS The old grandmother said, "I can take you down the mountain, but you must come back, because your mother and father do not love you." topic: lao nainai 'old grandmother’ entity: E Human, TP = 1 n topic: wo 'I’ entity: E Human, TP = 0 n topic: nimenk 'you’ entity: E Human, TP = 7 topic: nimen de baba, mamaj’p 'your mother and father’ entity: E Human, TP = 6 Da Hui Lang 099 Suoyi, zhe liang ge xiao nuhai ti le, liang therefore this two CLSF small girl carry PRF two lanzi de shuiguo, ganguo, he shiwu qu kan tamen de basket GEN fruit dry fruit and food go see 3pl GEN baba, mama. father mother Therefore, the two little girls carried two baskets of fruit—-dried fruit and food--and went to see their _mother and father. 123 Da Hui Lang 099 (con’t) topic: zhe liang xiao nuhaik 'the two little girls’ entity: E Human, TP = 5 Da Hui Lang 100 (EB) Dang tamen zou dao ziji de jia men de shihou ne, when 3pl walk to self GEN home gate RLV time PRTL tamen de baba, mama you lao you er, zuo zai 3pl GEN father mother have old have hungry sit at men kou, chi yidian sheng fan. gate mouth eat little remaining food When they walked to the gate of their home, their parents, old and hungry, were sitting in the gateway eating what little remaining food they had. topic: dang tamen zou dao ziji de jia men de shihou me 'when they walked to the gate of their home’ entity: IC (shihou, tamen zou)/SA: progression of time Abstract, TP = 0 138 Hui Lang 101 Liang ge xiao haizi mei you qu jiao ta baba, mama, two CLSF small child not have go call 33 father mother bu liang lan shiwu fang zai men kou, jiu zou no two basket food leave at gate mouth then leave 1e. PRF The two little girls did not call to their mother and father, but left the baskets of food at the gate and left. topic: liang ge xiao haizik 'the two little children’ entity: E Human, TP = 4 Da Hui Lang 102 (EB) Houlai ta baba, mama, yinwei bu xihuan zhe liang later 33 father mother because no like this two ge xiao hai, xinling shou le qianze; you CLSF small child soul suffer PRF guilt moreover mei you dongxi chi, jiu lian bing dai er de not have thing eat then two sick and hungry PRTL 124 Da Hui Lang 102 (con’t) siqu le. die PRF Later, because the mother and father did not love these two little girls, they suffered guilt, and since they did not have anything to eat, they became sick and hungry and died. topic: ta baba, mamaj 'the mother and father’ . 3 P entity: E Human, TP = 4 I)a Hui Lang-103 Ta de liang ge da jiejie jiu jia 1e nanren 3s GEN two CLSF big older sister then marry PRF man gen tamen de nanren zou le. with 3pl GEN man leave PRF Their two older sisters then married men and with them left. topic: ta de liang ge da jiejie 'the two big sisters’ entity: I (si ge nuerBN)/part to whole Human, TP = 1 Da Hui Lang 104 Zhe liang ge xiao nuhai gen lao nainai this two CLSF small girl with old grandmother yiqi chang da. together grow big These two little girls grew up together with the old grandmother. topic: zhe liang ge xiao nuhai gen lao nainai 'the two little girls with the old gran mother’ entity: E ‘ Human, TP = 1 13a Hui Lang 105 Tamen guo de feichang xinfu. 3pl live MNR very happily They lived very happily. topic: tamenk,o 'they’ entity: E Human, TP = 2 125 Da Hui Lang 106 Tamen yizhi zai shan li zhu, zhu de feichang 3pl all the time at hill in live live MNR very hao. happy All the time they lived in the hills, and lived very happily. topic: tamenk’o 'they’ entity: E Human, TP = 1 S I MAGUANG TEXT This text concerns the heroic efforts of Si Maguang. Once, as a child, one of his friends fell into a large water container. Relying on a cool head and quick thinking, he rescues the child. 81 Maguang 001 Si Maguang shi Song Chao hen chuming de zhengzhi Si Maguang is Song dynasty very known RLV political gaigejia. reformer Si Maguang is a well known political reformer or the Song Dynasty. topic: Si Maguangi entity: BN Human, TP = 1 Si Maguang 002 Ta zai xiaoshihou, you yi ci gen henduo xiao 33 at childhood have one situation with many small pengyou, nannan nunu de xiao pengyou zai changshang -friend boys girls RLV small friend at yard wan. play One time, when he was a child, many of his friends were Playing in the yard. tepic: tai entity: E Human, TP = 0 Si Maguang 003 Zuiwan de shihou, you yi ge xiaohai, hurang duo Playing RLV time have one CLSF child suddenly fall J'in shui gang li, yi ge da de shui into water 'container in one CLSF big RLV water gang li. Container in 126 127 Si Maguang 003 (con’t) While they were playing, one child suddenly fell into a large water container. topic: zuiwan de shihou 'while they were playing entity: IC (shihou, line 2)/SA: progression of time Abstract, TP = 0 Si Maguang 004 Zhe ge shui gang zhuang mang le shui. this CLSF water container contain full PRF water This water container was full of water. topic: zhe ge shui gangj 'the water container’ entity: E Inanimate, TP = 0 S i Maguang 005 Zhe xuduo xiaohai dou jingya le. this many child all surprise PRF This surprised all of the children. topic: zhe xuduo xiaohai douk 'all the children’ entity: E Human, TP = 3 Si Maguang 006 " Zenme bang? " what do "What should we do?" topic: ¢k entity: E II M Human, TP Si Maguang 007 "Zenme qu jiu zhe ge xiaohai?" what go save this CLSF child "How do we save this child?" topic: gk entity: ll g... Human, TP 128 Si Maguang 008 Si Maguang shuo, xiang le yi xiang, "Bu yao ji, wo Si Maguang say think PRF one think no want worry ls lai xiang ge bangfa." come think CLSF way Si Maguang said, after thinking awhile, "Don’t worry, I will come and find a solution." topic: Si Maguangi entity: E Human, TP = 6 topic: ¢k entity: E II 0 Human, TP Si Maguang 009 Ta zhao le yi zhao, zhaodao yi kuai shitou, yi 3s look PRF one look find one piece stone one kuai da shitou. piece big stone After looking around for a while, he found a large stone. topic: ' tai 'he’ entity: E Human, TP = 3 Si Maguang 010 Ta naqi zhe ge shitou, yong le li shuipo le 33 pick this CLSF stone use PRF strength break PRF na ge shui gang. that CLSF water container He picked up this stone, and used his strength to break the water container. topic: tai 'he’ entity: E Human, TP = 1 Si Maguang 011 Zhe yang, shui gang li de shui hen zirang this way water container in RLV water very naturally 129 Si Maguang 011 (con’t) jiu liu chu lai le. immediately flow set.out come PRF In this way, the water, that was in the container, liaturally flowed out. ‘topic: zhe yang 'this way’ (entity: I (line 10)/pr0position=situation Abstract, TP = 0 53:1. Maguang 012 INa. ge gang li diao xiaoqu de xiaohai de jiu ‘that CLSF container in fall into RLV child PRTL save le. (3R8 'The child that fell into the water container was saved. ‘topic: na ge gang li diao xiaoqu de xiaohail 'the child that fell into the water container’ entity: E - Human, TP = 0 .__s.~ fir" ‘- DUSHUREN TEXT The Dushuren text is about a student who needs to get to ‘tlie city before the city his servant carrying his books,- Bui;“the servant falls and the books scatter. takes to pick them up, Dushuren 001 ‘chu yi ci, yi 86 have one situation one CLSF student cheng. cxity One time , tx>pic: entity: ' you yi Ci I/SA: Abstract, TP = 0 DllSihuren 002 T8. dai le shutong yiqi 3&3 take PRF servant together go Clieng. City TRDgether with his servant, topic: tai ‘he’ entity: E Human, TP = 7 DLIE-huren 003 T8. dai le hen duo shu, :38 take PRF very many book shutong bei zai jian Servant carry at shoulder gates close for the night. With the two hurry for the city. In the time it the gates of the city close. dushuren, ta yao jin 3s want enter a student wanted to go to a city. 'one time’ a story must occur in time 9“: yiqi yao jin together want enter they went to the city. yong shengzi kun hao le you use rope tie well PRF by shang, yao jin cheng. on want enter city lie had many books, which were tied together with a rope find which his servant carried on his shoulder. ‘topic: tai 'he’ entity: E Human, TP = 6 130 131 Dushuren 004 Keshi, zhe ge ne, taiyang xiaoshan de shihou men but this CLSF PRTL sun set RLV time gate yao guan de cheng men yao guan de. want close PRTL city gate want close PRTL But by this time the sun was setting and the gates of the city were to be closed soon. topic: zhe ge ne 'this city’ entity: E Inanimate, TP = 2 Dushuren 005 Zai zhe ge shihou, jiyu yao zou, yao jin zhe at this CLSF time hurry want leave want enter this Be cheng. CLSF city At this time, they begin to hurry so that they can enter the city. tepic: zai zhe ge shihou 'at this time’ entity: E Abstract, TP = O DuShuren 006 . Zhe ge dushuren gen zhe ge xiaohai dou jiyu yao this CLSF student with this CLSF child all hurry want 5 in cheng. enter city Both the student and the child hurry to enter the city. topic: zhe ge dushuren gen zhe ge xiaohaii j 'the stu- dent and the child’ ’ entity: E Human, TP = 6 Dushuren 007 Teibei na ge xiaohai, ta you bei le shu, especially that CLSF child 38 also carry PRF book hen zhong, you jiyu yao zou. Very heavy also hurry want leave Especially the child, who was carrying the heavy books, also wanted to hurry. ‘ 132 Dushuren 007 (con’ t) topic: teibei na ge xiaohaij ’especially that child’ entity: E ' Human, TP = 6 Dushuren 008 Nianji you xiao. age also small He was also young. topic: nianji 'age’ entity: I (xiaohaiE)/SA: characters have age Abstract, TP = O Dushuren 009 Jieguo zai lu shang, shuai dao 1e. therefore at road on fall down PRF Therefore, he fell down on the road. tepic: zai lu shang ’on the road’ entity: I (line 7)/SA: people travel on roads Inanimate, TP = O DuShuren 010 Yi shuai dao le yihou, ta de shu wanchuan Soon fall down PRF after 33 GEN book completely San le. fall apart PRF Soon after he fell down, his books came apart. topic: yi shuai dao 1e yihou ‘soon after he fell down’ entity: I (line 9)/SA: progression of time Abstract, TP = 0 Dllshuren 011 Shengzi duan le. rope break PRF The rope broke. topic: shengzi 'rope’ entity: E Inanimate, TP = O 133 Dushuren 012 Jieguo deng duo na ge shu de, shutong ba zhe therefore wait until that CLSF book GEN servant DO this xie shu naqi lai, yao zhongxin kun. CLSF book pick up want again tie Therefore, he had to wait until the servant pick up the books and tied them together. topic: ¢i entity: E Human, TP = 2 Dushuren 013 Name na ge dushuren, gen zhe ge shutong like that that CLSF student with this CLSF servant Yiqi dao zhe ge cheng men de na ge together arrive this CLSF city gate RLV that CLSF Shihou, men yijin guan le. time gate already close PRF By the time the student and the servant arrived at the City, the gates were already closed. ’ ‘topic: na ge dushuren . . . de na ge shihou ’that stu- dent . . . that time’ «entity: IC (shihou, dushuren dao cheng)/SA: progression of time Abstract, TP = 0 HESHANG TEXT The Heshang text begins at an old temple in the moun- ‘téiinS where a monk lives by himself. Everyday he goes down the mountain to get water. Soon after, a second monk joins kiiln, and together they get the water. Then a third monk ar— Iéinves and troubles begin. The first monk thinks the other ‘ttvcb should get the water since he has done it by himself for :3<> long. The second monk complains of poor health and a need to pray all the time. The third monk claims he does not know his way around and may get lost or even break the Water container. They argue like this all day. When it Starts the rain, they think their troubles are solved-~some- <311