# Introduction We have to give a grammatical explanation of the term, left-dislocation. Left-dislocation is a marked word order, which derives from the canonical SVO word order of English (Ross, 1967). Traditionally, leftdislocation has been characterized in terms of the preposing a noun phrase from a proposition into the initial sentence position that is external to it, which contains a coreferential pronominal reference with the left-dislocated noun phrase (Ross, 1967). Examples are here: 1) Cathy, she is not a good friend herself. 2) That boy, he's supposed to be awe some. In ( 1) and (2), each initial noun phrase, 'Cathy' and 'that boy' is supposed from each proposition: the canonical word order of ( 1) is 'Cathy is not a good friend herself' and that of (2) is 'That boy's supposed to be awe some.' II. # Theoretical Background In general, many types of research on leftdislocation have shown that the function is to topicalize a referent (Halliday, 1967;Reinhart, 1981). However, these studies focus on the function of left-dislocation in one sentence rather than in a discourse context. Keenan-Ochs and Schieffelin (1976), who have examined the discourse function of left-dislocation, regard the usage of left-dislocation in discourse as a speaker's strategy in the communicative work. They propose that the speaker brings a referent into the foreground of the listener's consciousness by leftdislocation; then, the referent is usually not currently a center of attention; that is, it is not the current topic in the discourse context (p. 242). They indicate three discourse functions of left-dislocation: (i) to introduce a discourse-new referent, (ii) to reintroduce a referent into n this paper, we describe the discourse-pragmatic functions of left-dislocation in American English discourse and descriptively observe how the particular word order relates with the discourse context or the discourse unit; moreover, with a speaker's communicative intention. As for the term discoursepragmatics, in this paper, we follow the definition by Lambrecht (1994), in the state that "since discourse involves the use of sentences in communicative settings, such research is associated with the general area of pragmatics (p. 2)". In addition, in this paper, we put forward two terms, which are discourse-pragmatic function and discourse function. In this thesis, the analyses argue the discourse-pragmatic functions through the usages of left-dislocations as a speaker's intention, assistance, and goal in the communicative discourse context. On the other hand, the discourse function draws upon the phenomena of the work in the discourse context. # I Given the non-canonical structure of leftdislocation, this paper aims to investigate how this marked word order construction is used for the speaker's aim to achieve some goals which would not be achievable by using the unmarked counterparts in the discourse context. It is significant to clarify that we are not trying to analyze the grammatical derivation of left-dislocation but to examine why some noun phrases are preposed in the discourse context. To this end, we rely on the functional linguistic perspective approach and attempt to describe the use of left-dislocation in discourse has particular roles in the discourse construction or the discourse unit. Some recent studies which analyze linguistic forms in discourse contexts (Fox, 1993) show that discourse data is saliently distinguishable from monologic data. In this respect, this paper is based on the insights of functional and interactional-oriented studies that look at the functions of left-dislocation in the discourse. In the next section, we will introduce studies focused on discourse functions of left-dislocation (Keenan-Ochs & Schieffelin, 1976; Ono & Thompson, 1994;Gundel, 1985;Prince, 1985Prince, , 1997)). the discourse when the referent is not in the foreground of the speaker's and hearer's minds, and (iii) to give an emphasis upon a referent. They note that the former two functions, (i) and (ii), are functions in the use of leftdislocations; on the contrary, the latter function, (iii), is a minor. Each discourse function is shown in the following examples; 3) 1. K: Yeah// Yeah! No matter how old// you are 2. L: Yeah. Mh hm 3. Parents don't understand. 4. But all grownups w-they do it to kids. 5. Whether they're your own or not. (Keenan-Ochs & Schieffelin, 1976, p. 243) 4) 1. K: An' I got a red sweater, an' a white one, an' other 2. sweaters, you know, 3. And uh my sister loves borrowing 4. my sweaters because they're pullovers, you know, 5. she c'n wear a blouse under'em 6. an' she thinks "Well this is great" 7. An' so my red sweater, I haven't seen it 8. Since I got it. (Keenan-Ochs & Schieffelin, 1976, p. 243) 5) 1. L: T'know some of 'em are darmn tall and goodlooking 2. they could pass for (t)-nineteen.// 3. A twelve year old guy comes over 4. I say who's y-older brother is he? 5. He's not he's in the A7. 6. R: But they don't-7. But they don't have a brain to go with it hehhh 8. L: These kids I don't believe it they're six foot. (Keenan-Ochs & Schieffelin, 1976, p. 245-6) In example (3), the left-dislocation at line 4 introduces a discourse-new referent, 'all grownups.' The precious topic is about the speaker's parents (Keenan-Ochs & Schieffelin, 1976, p. 243). In example (4), the left-dislocation at line 7 reintroduces a referent, 'my red sweater,' into the discourse. The referent appeared at line 1; however, it is not a topic lines 3 to 6. The main is 'my sister' in the discourse context (Keenan-Ochs & Schieffelin, 1976, p. 243). The left-dislocations in these two examples function to foreground a referent; but the left-dislocation at 8 in example (5) does not. The left-dislocated referent, 'these kids', is in the foreground of the speaker's and hearer's minds in the preceding discourse context. Keenan-Ochs and Schieffelin (1976, p. 245) state that "the speaker is using the basic function of focusing the listener's attention on some referent to amplify the attention paid to some referent under discussion." They report this type is a minor focus function of left-dislocation, and it occupies 6.6 percent of the total left-dislocations in their data. From a similar perspective, Ono & Thompson (1994) suggest that the discourse function of left-dislocation is particular for establishing or tracking a referent in the discourse context. The characteristic of the left-dislocated referent or noun phrase is belonging to Keenan and Schieffelin (1976), Ono & Thompson (1994), Gundel (1985), and Prince (1985Prince ( , 1997)). According to Keenan-Ochs and Schieffelin, the left-dislocated referent typically seems to be an entity known to or knowable by the hearer from a non-verbal context or from prior background experiences; and it is an entity that the hearer can identify or recognize a referent of representation (p. 242). Similarly, Gundel (1985) and Prince (1985Prince ( , 1997) also point out that the possibility of an appearance of left-dislocation depends on a speaker's familiarity with a referent in the discourse. In other words, a referent which a speaker refers to by left-dislocation has a unique referent that the hearer can identify from previous utterances or extra-linguistic contexts, as for the formal features of left-dislocated noun phrases, Ono & Thompson (1994) note that they are defined as 100 percent fully specified noun phrases such as proper nouns or noun phrases with determiners or genitive modifiers (p. 410). In this study, we mainly reexamine the discourse functions of left-dislocation suggested by Keenan-Ochs & Schieffelin (1976) and Ono & Thompson (1994), and descriptively demonstrate it. As for the descriptive approach, we use a measurement proposed by Givón (1983). Therefore, it is accountable for in the next chapter. We then demonstrate how the left-dislocations in our data work in the discourse context and the discourse unit, based on the calculation. Next, we investigate the definiteness of the leftdislocated referents in our data that are identifiable for the hearer in the discourse context. Finally, we characterize the discoursepragmatic functions of the left-dislocations. Then, we attempt to simplify the speaker's communicative intention in the usages of left-dislocation in the discourse context. # III. # Data and Methodology The data for this study consists of three corpora of the Santa Barbara Corpus of Spoken American English. All of them are face-to-face conversations: one corpus is that over dinner in a private home among four male speakers, and they are friends. They mainly talk about the dance called the lambada and about a party. Another corpus is the face-to-face conversation with cooking dinner in the kitchen among three speakers. One of the three speakers is female, and the two are male; and they are friends. They mainly talk about the decay of nature or their personal stories. Another corpus is a face-to-face conversation at the living room among two speakers, and they are female friends. They mainly talk about their personal stories or a party to which the host invited one speaker. My database consists of 77 minutes of conversation transcribed into intonation units, of which there are 4633 (cf. Du Bois, Schuetze-Coburn, Cumming, & Paolino, 1993; Du Bois & Schuetze-Coburn, 1993). An intonation unit refers to a spurt of talk produced under a coherent intonation contour, often is connected to a pause. For the present study, we apply two calculations proposed by Givón (1983): one calculation is a "referential distance (look back)," and the other one is a "topic persistence." The calculation of referential distance shows the gap between the previous occurrence of a referent or a topic in the discourse and its current occurrence in a clause, significant for a particular grammatical coding device. The gap is obvious in the number of clauses to the left of the present occurrence of the grammatical coding device in the discourse context (p. 13). The test of referential distance will clarify whether the left-dislocated noun phrase in our data appears to introduce or reintroduce a referent in the discourse context. It also shows whether the left-dislocated referent is topical in the previous discourse context. Givón (1983) explains that the calculation of topic persistence, on the other hand, looks at the subsequent discourse. It shows the reflection of a topic's importance in the discourse context and the speaker's topical intention (p. 15). He says that "more important discourse topics appear more frequently in the following discourse (p. 15)." The persistence is measured in terms of the number of clauses to the right from a referent's present occurrence (p. 15). The test will clarify the topicality of the leftdislocated noun phrase in our data in the succeeding discourse context. For this study thus, the unit of analysis is a clause. Our database consists of 1753 clauses; these clauses are analytical. Moreover, we will measure not only the number of clauses but also the number of pronominal references which are coreferential with a left-dislocated noun phrase, and nominal references which refer to the same referent with a leftdislocated noun phrase because the measurement more clearly shows the referential distance and the topic persistence of left-dislocation in our data. As for the definition of left-dislocation in our data, we draw upon that by Ono & Thompson (1994): (a) the left-dislocated noun phrase has continuing intonation contours which signaled by a comma, not by a period in the transcription, (b) the left-dislocation causes to turn-taking by a speaker or a hearer between the left-dislocated noun phrase and the proposition, and (c) the left-dislocated noun phrase does not appear as a backchannel in the discourse context. Moreover, of course, we do not regard vocative as a left-dislocated noun phrase. # IV. # Analysis and Results In our data, the appearance of left-dislocation is 20 cases in the total. The total percentage of the total number of clauses (1753 clauses) is 1 %. We might say these are the limited ones. In this chapter, we analyze the discourse functions of the left-dislocations in the following sections, based on Givón's (1983) calculations. And we show the results of the referential distance and the topic persistence in the discourse context, in addition, to the forms and the stress of the left-dislocated referents. # a) Discourse Function Left-dislocations occur to introduce or reintroduce a referent on story-telling discourse contexts. In other words, the speakers use them to foreground a referent or establish a referent as a topic in a discourse context; however, in fact, some leftdislocations in discourse contexts seem to work in different ways in respect of foregrounding a referent or managing a discourse. It is for this reason that leftdislocations in our data have various characteristics in the referential distance in the preceding discourse context and the topic persistence in the succeeding discourse context. Each also interacts with the discourse functions of the left-dislocations. We divide the discourse functions of the leftdislocations in our data into four types: (1) to reintroduce a referent (hereafter Type 1), (2) to rephrase a referent (hereafter Type 2), (3) to characterize a referent (hereafter Type 3), and (4) to introduce a referent (hereafter Type 4). Type 1 is the discourse function of the left-dislocations in our data because it occupies 50 percent of the total; Type 2 is 35 percent, Type 3 is 5 percent, and Type 4 is 10 percent. Then, the left-dislocations in Type 2 do not have a clausal gap between a referent and a left-dislocated referent which refers to a referent. The left-dislocated referent refers to the immediate previous referent. In other words, a topic in the discourse context is not particular by the left-dislocation in contrast with the leftdislocations in Type 1, which sets it apart as changing a theme in the discourse context. In Type 2, the topic As for each characteristic of the referential distances, the topic persistence, the forms, and the stress of the left-dislocations, the left-dislocations in Type 1 have clausal gaps in the first introduction of referent into the discourse by the left-dislocation. In Type 1, the topic persistence is continuous in the left dislocations succeeding discourse context. Further, the forms of the left-dislocated noun phrases in Type 1 are proper nouns or nouns with the genitive modifiers. A lot of the left-dislocated noun phrases have the stress. discourse context, and this feature is also as opposed to that of Type 1. And the forms of the left-dislocated noun phrases in Type 2 are characterized by the following: the nouns with a demonstrative pronoun, 'that,' or with a definite article, 'the.' In addition, most of the left-dislocated noun phrases have the stress. Next, the left-dislocation in Type 3 does not have a clausal gap between a referent and the leftdislocated referent, which refers to the referent; and the characteristic of the referential distance seems to be similar to that in Type 2. However, they differ in the topicality of the referent referred to by the left-dislocation in the previous discourse context. The left-dislocation in Type 2 refers to a referent that has a high topicality in the previous stage, and the referent referred to by the left-dislocation does not continue as a topic in the succeeding discourse context. The left-dislocation in Type 3, on the other hand, refers to a referent that is not a topic in the previous discourse context, and the referent referred to by the left-dislocation continues as a topic persistence is lower than of in Type 1. As for the form of the left-dislocated noun phrase in Type 3, it is the noun with a demonstrative pronoun, 'this.' The leftdislocated referent has weak stress. # i. Type 1: Function to reintroduce a referent Here, we examine Type 1. The left-dislocation occurs to reintroduce a referent into the discourse context, where the speaker introduced the referent at the previous discourse; however, there is only one reference that refers to the referent at the context. The referent has not been a topic, but another referent has been a focus as a topic. Left-dislocations occur to focus on the referent as the topic in the discourse context. The speaker brings a referent into the foreground of the listener's consciousness by the left-dislocation. The following example (6) shows the case where the leftdislocation appears for the reintroduction of referent, which is not a central topic in the previous discourse context. Finally, the left-dislocations in Type 4, of course, do not have a clausal gap between a referent and the left-dislocated referent because the discourse function is to introduce a discourse-new referent. The topic persistence in the following discourse context is continuous, while it may not mean a topic persistence but rather a thematic structure. The difference between topic persistence and thematic persistence would be significant in the number of referential markings in the succeeding discourse context, described by the examples in the later section. As for the forms of the left-dislocated noun phrases, they are indefinite. And they have weak stress. We exemplify each type of the left-dislocations from our data in the following subsections. In this context, the speaker has told a story about which her acquaintance, 'Mike,' is the person who stole Hector's radio (line 11-12). 'Mike' and 'Hector's radio' are new in the discourse context. We can find the left-dislocation at 42-43: 'Hector's radio, it was broken.' The referent is reintroduced into the discourse because it has not been the center of the topic. The speaker has talked about 'Mike' from the line 11 to 38: 'he (=Mike) was a cocaine addict' at line 17, 'he (=Mike)'s talking about' at line 18, 'he (=Mike)'s gonna show us' at line 22, 'he (=Mike)'s actually pretty intelligent' at line 36, and 'he (=Mike) just uneducated' at line 37-38. The referent, 'Mike' is the main topic there; 'Hector's radio' is not. Accordingly, there is no an identified referent, 'Hector's radio' at line 13 to 41. After the interruptions such as 'so', 'um', and 'yeah' at line 39 to 41, the speaker has started talking about the referent, 'Hector's radio' at line 42. We can say that the function of leftdislocation is to reintroduce the referent into the discourse context. In this example, there are clausal gaps between the previous utterance at line 12 and the current one at line 42, where it contains nine clauses. # Volume XXII Issue II Version I # take it The left-dislocation is used to foreground the referent in the background of the hearer's consciousness and to make the topic-marking. The speaker has talked about 'Hector's radio' in the succeeding contexts at line 42 to 68: for example, 'we were gonna take it (=Hector's radio) out and send it (=Hector's radio) back to the factory' at line 44 to 45, 'he (=Hector) never got his radio' at line 58, and 'he (=Hector)'s having bad luck with that car' at line 68. There are ten pronominal or nominal referents that refer to or presuppose 'Hector's radio' after the leftdislocation at line 42-43. It is represented: 'it, radio, he, his radio, that, and we (because Hector is Alina's husband)'. There are 15 pronominal and nominal references which refer to the left-dislocated referent among 12 clauses. Thus, we see that the function can maintain the topic persistence. # ii. Type 2: Function to rephrase a referent The left-dislocations in Type 2 occur to rephrase a previous topical referent, and it is in the foreground of the listener's consciousness. The discourse function gives the speaker special emphasis or comment upon a particular entity mentioned in the prior utterance. Moreover, the topic persistence is not particular in the succeeding discourse context. Characteristically, the speakers use the left-dislocation when they finish talking a story about the referent. We found 7 cases or 35 percent of this type. 37 In this context, the speakers, Alina and Lenore, have talked about one girl who has been at a party to which she has gone at line 1 to 43. The first mention occurs as 'this young chick' in line 2. The leftdislocation is at lines 40 to 41: 'that stupid little bitch, she just married daddy to take care of her.' The leftdislocated referent refers to the referent, 'this young chick.' The left-dislocated referent is argued as the topical thing when the speaker uses the left-dislocation. There are 11 pronominal references in the succeeding discourse context of the nominal reference at line 2. The speaker gives a comment on the girl. In addition, in line 43, the speaker starts talking about another topic. Here, of course, it should be sure that the pronominal reference, 'they,' refers to the girl in line 43. However, the center topic or story in the following discourse is not focused on only the girl, but rather on the lifestyle of the older guy and the girl of line 1-2. In this sense, the topic change is noticeable in the discourse context. In summary, we can say that the left-dislocation is used to rephrase the previous referent, that is, to give a final comment on a story and to give an emphasis upon the referent. iii. Type 3: Function to characterize a referent 8) 1. ALINA: Well I didn't get along with !Dennis at all, 2. he was a jerk. 3 24. and he was real bright, In this context, the speaker, Alina, is talking about the event at a party she visited. There are many persons in this discourse context: 'Dennis' at line 1, 'Spargo' at lines 5 and 21-22, and 'two paddlers' at line 9. The topic changes one after another in the discourse context. The left-dislocated referent at line 10-11 refers to the immediate previous utterance at line 9. In this example, the speaker characterized the referent by the left-dislocation. As for the topic persistence, the left-dislocated referent, 'these guys,' is referred to by pronominal references in the following context: 'each one' at line 16 and 'they' at line 20. We can say that the topic persistence is lower than that of referent in Type 1. Moreover, the left-dislocated noun phrase occurs with the demonstrative 'this. ' The noun phrase has weak stress. # iv. Type 4: Function to introduce a referent In this section, finally, we demonstrate the discourse function of the left-dislocations in Type 4. The left-dislocated referent occurs in the discourse context. In other words, the left-dislocations are significant to introduce a referent. We found 2 cases or 10 percent in this type of total. 9) 1. MARILYN: ? Mhm. 2. ? Yeah. 3. ? little lemons from the tree, 4. they're still kinda yucky. 5. .. <@ You know @>, 6. (H) we came back from a, 7. ? we had to go to ? the Ritz Carlton, 8. , 9. .. for a .. event. In the context, Marilyn has talked about other topics. In this example, the left-dislocated referent at line 3, 'little lemons from the tree' appears into the discourse context. The left-dislocated referent, 'little lemons,' is not directly referred to by the speaker in the following discourse. There is not a nominal or pronominal reference. The left-dislocated referent is marked as the theme in the succeeding discourse. We might say that the speakers use the function to establish or track a referent. It is a discourse-new referent. # 10. In this section, we examine the left-dislocation in Type 3. The left-dislocated referent refers to an immediate previous referent; however, the referent is not topical in the discourse context. The speaker uses the left-dislocation to characterize the utterance by themselves. # b) Result We have examined the discourse functions of the left-dislocations in our data in the preceding section, and proposed four discourse functions: (1) to reintroduce a referent (50%), (2) to rephrase a referent (35%), (3) to characterize a referent (5%), and (4) to introduce a referent (10%). Each characteristic has been mainly significant by two measurements of referential distances and topic persistence. Table 1 indicates each average of the referential distance in four discourse functions of the leftdislocations; on the other hand, Table 2 shows each of the topic persistence in four discourse functions of the left-dislocations. Referential distances show the clausal gap between the previous occurrence of a referent or a topic in the discourse and the left-dislocated referent. In addition, we have analyzed the number of references. The result in Table 1 manifests that the leftdislocated referents which have the function of reintroducing a referent (Type 1) has clausal gaps on average, which marks 30 clauses. Table 2 shows each persistence of the leftdislocated referents in the left dislocations following discourse context, which was measurable by the number of the clauses and references. We can descriptively analyze the functions in our data by the measurements which show referential V. # Discussion In the preceding chapter, we have analyzed the discourse functions of the left-dislocation in our data. Then, they have been divided into the following four types: (Type 1) to reintroduce a referent; (Type 2) to rephrase a referent; (Type 3) to characterize a referent; and (Type 4) to introduce a referent. In this chapter, we will, first of all, examine the identifiability, that is, the definiteness of the leftdislocated referents in the discourse context. Next, we will point out that the four discourse functions which we have shown can be further generalized down to the two discourse-pragmatic ones in using the left-dislocations. The two discourse-pragmatic applications interact with the speaker's goal, which s/he attempts to achieve by using the left-dislocation in the discourse context. Of course, they are particular by basing on the results of the referential distance and the topic persistence of leftdislocation, which we have clarified in the preceding chapter. Then, we here will focus on the characteristic of the left-dislocations in the thematic paragraph rather distances and the topic persistence. Next, we focus on the speaker's intention in the usages of the functions; we also investigate the roles of the left-dislocations in the discourse or thematic unit. Each result concerning forms and stress of them is finally visible in Tables 3 and 4. # Average in clauses Average in reference Average in clauses Average in reference than the discourse context which we have analyzed. In addition, we show the stress pragmatically has the markedness and clearly indicates the boundary of the thematic paragraph in the discourse unit. # a) Definiteness We, thus, try to descriptively explain the correlation between the form and the discourse functions of the left-dislocations in our data and to give an explanation of which all the left-dislocated referents in our data are familiar and identifiable for the speaker and the hearer in the discourse context even though they are introduced in the discourse context, as it is certain in the usages of left-dislocated referents in Type 4. The leftdislocated noun phrases at Type 4 were indefinite; on the other hand, the other left-dislocated noun phrases, particular in Type 1, 2, and 3 were definite. That is to say, the left-dislocated referents in Type 1, 2, 3 seem to be familiar or identifiable referents for the speakers and the hearers; however, that in Type 4 does not seem to be. We shall now focus on the definition of definiteness. Chafe (1976) defines definiteness as a speaker's assumption if the hearer knows or can identify a referent the speaker refers to or has in mind in communicative situations. In other words, it means that the hearer correctly picks up the referent that the speaker refers to, which is identifiability (p. 39). Similarly, Givón (1983) says that definiteness shows that a hearer can uniquely identify a referent referred to by the speaker. The speaker can assume that the referent is familiar to the hearer (p. 10). Chafe (1976) points out that the identifiable referent is figured out such nouns as proper nouns, common nouns, generic nouns, and nouns with the determiners like 'this' or 'that.' Moreover he clears that nouns are distinguishable as direct labels to particular referents; and the way to construct definiteness is a base on a speaker's prior mention in the discourse context. Thus, contexts or scenes are special for definiteness (p. 39). Let us now expand these arguments into the left-dislocated referents in our data. # Shibata (1975), who observes lexical concreteness, refers to several examinations on proper nouns. He explores that a proper noun is limitedly a reference that refers to a particular person among friends. However, it is essentially able to apply to or refer to any person. On the contrary, in the case of semantic generalization of proper nouns, the proper nouns refer to a characteristic of a famous person by using a famous person's name (p. 48). For example, it is expressed as 'he is (like) Bill Gates.' We all notice that Bill Gates is identifiable as a rich man. Next, we observe the correlation between the left-dislocated nouns with the determiner, 'that' or the definite article, 'the' and the discourse function to rephrase a referent (Type 2). The left-dislocations occurred to emphasize a topical referent in the immediate previous discourse context. Shibata (1975) notes that demonstrative or personal pronouns cannot be correctly understood by a hearer until they appear in a context (p. 41). In that sense, the left-dislocated referents seem to be anaphoric usages in the discourse context. As for anaphoric usages, Givón (2001, p. 196-199) explains that an anaphoric referent has to refer to a topical referent in the previous discourse context. An anaphoric reference requires "functional transparency" (p. 196). If an anaphoric reference refers to a non-topical referent in the discourse, it is difficult for the hearer to identify the referent to which the anaphoric referent refers. As for definite/indefinite referents in discourse, Givón (1983) also addresses that a hearer has to open a new file in their mind when a topic is particular by an indefinite noun phrase, it may be difficult for a hearer to identify the referent in such case as that there are clausal gaps between the definite referent and the First, we examine the relationship between the left-dislocated proper nouns and the discourse function to reintroduce a referent (Type 1); and it is the most frequently used in our data. All the forms were judgeable as proper nouns. The characteristic of the discourse function was characteristical that there were clausal gaps between the first mention of a referent and the second mention by the left-dislocation (see Table 1). It was 30 clauses on average. We may say that the speaker assumes that the hearer can correctly identify or pick up the left-dislocated referent even if the speaker reintroduces into the discourse context after 30 clauses from the first mention. In this section, we examine the definiteness of the forms of the left-dislocated referents which we have seen, as we have introduced, Keenan-Ochs & Schieffelin (1976), Ono & Thompson (1994), Gundel (1985), and Prince (1985Prince ( , 1997) ) note about the form of a left-dislocated noun phrase or the characteristic of a leftdislocated referent. For example, they are proper nouns, nouns with determiners such as 'this' or genitive modifiers (Ono & Thompson, 1994). Besides, it is a familiar or identifiable referent for the speaker and hearer (Gundel, 1985;Prince, 1985Prince, , 1997). However, they do not clarify the relation between a left-dislocated noun phrase and the discourse function. They only say that the discourse function of left-dislocation is distinguishable as establishing or tracking an identifiable referent for the speaker and the hearer in the discourse. The left-dislocated proper nouns used in our data refer to a particular person among friends, so it is identified as direct labels. We now can say that concrete proper nouns seem appropriate for the left dislocations discourse functions to reintroduce a referent because the hearer can identify the referent, that is, the familiar name. previous mention to which the referent refers. A hearer can identify a topic in such case as a short clausal gap between them (p. 11). Each left-dislocated anaphoric or definite referent with the demonstrative, 'that' or the definite article, 'the,' which is particular to give a special emphasis, referred to a topical referent in the preceding discourse context. If it was significant to reintroduce a referent, as seen in Type 1, it would be difficult for the hearer to correctly identify the referent, which the speaker refers to by the left-dislocation. Therefore, we may say that the anaphoric or definite characteristics of the left-dislocated referents related with the leftdislocated discourse function to give an emphasis or comment upon the immediate previous referent. Thirdly, we examine the correlation between the noun phrase with the demonstrative, 'this' and the left dislocations discourse function to characterize a referent (Type 3). This discourse function was distinguishable as follows: the left-dislocated referent referred to a referent in the immediately preceding utterance. And the leftdislocated referent, 'these guys,' was used to characterize the immediate previous referent, 'two paddlers. ' We can see that this case is also defined as an anaphoric usage which seems to be similar to the left-dislocated noun phrases with the demonstrative, 'that' and the definite article, 'the,' as we have seen. However, there is one significant difference in the demonstrative, 'that' and 'this'. Gundel, Hedberg, and Zacharski (1993) mention that the noun phrase with the demonstrative determiner 'this' has to be "speaker-activated, by virtue of having been introduced by the speaker or otherwise included in the speaker's context space (p. 279)". Observe the following examples: 11) B: My neighbor has a dog. This/that dog kept me awake last night. (Gundel, Hedberg, and Zacharski, 1993, p. 279) In these examples, the referent, 'the neighbor's dog,' which is referred to by 'this dog' in (10), is uttered by the speaker A; on the other hand, in (11) speaker B introduces the referent, 'a dog' which is clear as 'this dog' or 'that dog. ' Gundel, Hedberg, & Zacharski (1993, p. 279) draw attention to the usage of demonstrative determiner, 'this' in (10), is not appropriate. Thus, we can recognize that the noun phrase with 'this' is characterized by the speaker-oriented usage and anaphoric usage. The left-dislocated noun phrase with the demonstrative determiner, 'this' in our data, referred to a referent introduced by the speaker in the immediate previous utterance. It shows the anaphoric characteristic of 'this.' In addition, the speaker characterized the referent which the speaker introduced into the discourse by the left-dislocation; that is, it is clear as the speaker-activated referent. Thus, we can notice that the feature of the demonstrative determiner, 'this,' interacts with the discourse function of the leftdislocation. Finally, we explain the relationship between the left-dislocated indefinite noun phrase and the discourse function to introduce a referent (Type 4). These leftdislocations occurred to introduce a referent into the discourse for the first time, and all the forms of the leftdislocated noun phrases were indefinite. It seems to be a natural correlation between the discourse function and the form because it is new information in the discourse; however, it contradicts the examination proposed by Gundel (1985) and Prince (1985Prince ( , 1997)). They say that a left-dislocated referent is familiar or identifiable for a hearer. Is each indefinite referent of left-dislocated in our data non-identifiable for each hearer in the discourse context? We regard the indefinite noun phrases of leftdislocated as identifiable referents for the hearer in the discourse context because they are knowable by the hearer from the non-verbal context of the utterance, or "inferable" from the previous one (Prince, 1981). As for the inferable information in discourse contexts, Prince puts forward that the information status in discourse does not depend on the only binary distinction of whether it is given or new information. In addition, she proposes that new information is not always new information for a hearer (p. 236). The new information is distinguishable into two types; whether it is inferable or not from the previous utterance or the discourse situation. We here point out that the left-dislocated indefinite referents, which introduce a referent, are inferable. The one is inferable from the discourse situation; for example, the left-dislocated referent, 'little lemons from the tree,' is defined as the inferable referent from such extra-linguistic context as dinner time. The other inferable referent from the previous utterance may be focused on the inferable referent from the topic about an event. Thus, we can say that the left-dislocated indefinite referents which function as introducing a new referent are not identifiable as discourse-new referents but rather identifiable referents. The fact reveals that all left-dislocated referents in our data are familiar or identifiable for the hearer in the discourse context, even if they are indefinite. We have given more descriptive explanations regarding the familiarity and the identifiability of the left-dislocated referent than the simple observation suggested by Keenan-Ochs & Schieffelin (1976), Gundel (1985), Prince (1985Prince ( , 1997)), and Ono & Thompson (1994). In the following section, we shall examine that the degree Year 2022 G Discourse-Pragmatic Functions for Applications to Shared and Thematic Information Structures 10) A: Have you seen the neighbor's dog? (Gundel, Hedberg, and Zacharski, 1993, p. 279) B: Yes, and ? this dog kept me awake last night. that dog of the identifiability of the left-dislocated referents closely links with the discourse-pragmatic functions. In other words, it sets out the problem "referential accessibility" or "predictability" of the referent in the discourse context, which is noticeable by Givón (1992). # b) Discourse pragmatic functions In this section, we suggest that the four discourse functions of the left-dislocations we have analyzed seem to be further generalized down to the two discourse-pragmatic functions. The one discoursepragmatic function shows the function to topicalize a non-topical referent in the discourse context; on the other hand, the other discourse-pragmatic function shows the function to give an emphasis upon a topical referent. The former function indicates Type 1, 3, and 4, which we have seen, and the latter function indicates Type 2, which we have seen. The two discoursepragmatic functions of the left-dislocations are different. Each discourse-pragmatic function is particular by not only the differences in the referential distance and the topic persistence but also the differences in the anaphoric feature, which we have examined in the preceding section. In other words, the predictability of the left-dislocated referent in the discourse context relates with the discourse-pragmatic functions. Of course, all left-dislocated referents in our data are identifiable for the hearer, as we have observed; however, the degree of identifiability seems to be different in respect of the predictability of a referent. First, as for the topic persistence, the leftdislocations in Type 1, 3, and 4 have the topic persistence in the following discourse context; on the contrary, the left-dislocations in Type 2 do not have it. Moreover, each left-dislocation in Type 2 refers to the previous topical referent. That is, the referential distance is nothing. We can be fairly certain that the two kinds of the left-dislocations differ in the goal which the speaker attempts to achieve in the discourse context: in some cases, the speaker tries to foreground a non-topical referent in the discourse context to topicalize the referent in the succeeding discourse context; in other cases, the speaker would give an emphasis on a topical referent. In the former case, the speaker uses the leftdislocation in the opening of the story-telling discourse context; on the other hand, in the latter case, the speaker uses the left-dislocation in the last of the storytelling discourse context. The fact seems to relate to the thematic paragraph in each discourse unit. Let us discuss the relationship between the thematic paragraph and the discourse-pragmatic functions of the lest-dislocations. Later we shall try to give an account of the correlation of the predictability of the left-dislocated referents and the discoursepragmatic applications. Givón (1983) demonstrates that a topic-marking device is related with the construction of thematic paragraphs in the discourse unit (p. 6). Moreover, it is coded to a grammar syntax (p. 9). To begin with, regarding a thematic paragraph and discourse unit, he explains that a thematic paragraph is a focus by chains of clauses; further, the thematic paragraph is composite into larger discourse units such as sections, chapters, parts, or stories (p. 7). The thematic paragraph and discourse unit are strongly distinguishable as macro organizations in discourse; on the other hand, a clause chain is completely particular as a micro organization. As for the chains of clauses, Givón (1983, p. 9) suggests that there are three major types of the topicmarking device within thematic paragraphs: (a) chain initial topic, (b) chain medial topic, and (c) chain final topic. The characteristics of the three types are noteworthy below: a) Chain initial topic: (1) Characteristically a newly-introduced, newlychanged or newly-returned topic; thus (2) Characteristically a discontinuous topic in terms of the preceding discourse context; but (3) Potentially -if an important topic -a rather persistent topic in terms of the succeeding discourse context. b) Chain medial topic: (1) Characteristically a continuing/continuous topic in terms of the preceding discourse context; and also (2) Characteristically persistent -but not maximally so -in terms of the succeeding discourse context, even when an important topic. c) Chain final topic: (1) Characteristically a continuing/continuous topic in terms of the preceding discourse context; but (2) Characteristically non-persistent topic in terms of the succeeding discourse context, even if an important topic (Givón, 1983, p. 9) Volume XXII Issue II Version I Here, we regard two of three types, (a) and (c), as the discourse-pragmatic functions of the leftdislocations in our data: the initial chain topic seems to be coded to the left-dislocation which functions as topicalizing a referent in the discourse context; on the other hand, the final chain topic seems to be coded to the left-dislocation, which is distinguishable as giving an emphasis upon a topical referent. In addition, we assume that the chain medial topic is connected to the pronominal references in subjects' or objects' positions. Then, we now attempt to observe the chain medial topic device of the subject and the object in our data before turning to the closer examination of the two functions of the left-dislocations in the thematic paragraph. It would be helpful to describe our main task. We here provided the referential forms of the subject and the object constituents in our data (1753 clauses): 1690 noun phrases and pronouns in the subject positions, in addition, 866 noun phrases and pronouns in the object positions are examined. Tables 5 and 6 show each result. In both Table 5 and 6, the main point to note is to set out the percentage of each pronominal reference. Table 5 tells us that the pronominal ones account for 88 percent of the total of the subject constituents in our data; moreover, Table 6 shows that the pronominal references account for 26 percent of total of the object constituents in our data. Surely, the pronominal ones in the subject positions command an absolute majority in the subject constituents. In general, pronominal referents refer to given information or active information in the discourse context in English (Chafe, 1987(Chafe, , 1994)). Although the fuller study of the problems about subjects and objects, or given and new information lies outside the scope of this paper, in short, we regard each pronominal reference in the subjects and the objects as a continuing topic in the discourse context. That is, they code the medial chain topic. Let us now look at the problem of the leftdislocations and the thematic paragraph. As mentioned above, the left-dislocations code to the chain initial topic or the chain final topic of the thematic paragraph. And the pronominal references in the subjects and the objects code the chain medial topic. We are now able to see that the left-dislocated referent which codes the chain initial topic causes the referential and the thematic "disruption"; on the other hand, the left-dislocated referent that codes the chain final topic maintains the referential and the thematic "continuity" in the discourse, but not in the succeeding discourse context. Then, we have one question: can the speaker use left-dislocated referents as the topicmarking devices if they are identifiable for the hearer? We here that the left-dislocations which mark the chain initial topic and function as topicalizing a nontopical referent have an engagement in the shared information for the speaker and the hearer; on the contrary, the left-dislocations which mark the chain final topic of giving a special emphasis upon a topical referent have an argument on the textually thematic information in the discourse context. Tomlin (1986) notes the difference between shared information and thematic information: (a) Shared information: information in an expression is shared to the extent the speaker assumes the hearer is able to identify the referent of the expression. (b) Thematic information: information in an expression is thematic to the extent the speaker assumes the hearer attends to the referent of the expression. (Tomlin, 1986, p. 39) Before explaining the overt reasons for our hypothesis, we here provide the remarks on the predictability of a referent in discourse contexts, which is weighty by Givón (1992). Givón (1992, p. 12) highlights that "any information transacted in discourse has a certain level of predictability, coherence, or accessibility vis-à-vis its context." In addition, he points out "generic ("cultural") permanent memory" and "textural ("discourse") episodic memory" regarding the predictability or the accessibility of information in discourse, namely, regarding the shared contexts in the searches of information in the speaker's and hearer's present memory. Based on Givón's and Tomlin's ideas, we may regard that most of the notices of the left-dislocated referents coded to the clausal initial topics exist in the generic permanent memory of the speaker and the hearer; on the other hand, most of the information of the left-dislocated referents coded to the clausal final topics exist in the textual (discourse) episodic memory in the discourse context. As we investigated in the previous section, for example, the left-dislocated referents, which function as reintroducing a referent, are shed light on the proper nouns as the direct label to a particular person, such as the speaker's family. The searches depend on the shared concrete information between the speaker and the hearer rather than on the discourse context. That is, it exists in permanent memory. Therefore, when the referent has a topicality by the left-dislocation in the chain initial position, the hearer can easily understand the beginning of the referent which the speaker refers to and does not have many surprises because it is the certainly shared information. On the other hand, the left-dislocated referents which occur in Types are to set anaphoric referents. For example, it is made upon of such noun phrases as 'that boy.' They overtly exist in the textual episodic memory of the speaker and the hearer. Actually, the referents are introduced as the leftdislocated referent in the final of the story-telling discourse context; that is, it is cleared up the chain final topic. If a referent is introduced as a left-dislocated referent as the chain first topic in thematic paragraphs, the speaker cannot understand the referent which the speaker refers to. In this sense, we can summarize that the left-dislocation as the chain initial topic device follows the constraint of which the referent is identifiable and shared information in the discourse context; in contrast, the left-dislocation as the chain final topic device follows the constraint of which the referent is thematic information in the discourse context. In fact, the speakers use the two different discourse-pragmatic functions of the left-dislocations in our data. The choice is a base on the speaker's communicative intention and the feature of the information in the discourse context. Finally, we note the characteristic of the stress of the left-dislocations. As we have analyzed, many leftdislocated referents have stress. In general, English has a normal postverbal stress either on a direct object or on some other elements (Harlig & Bordovi-Harlig, 1988. P. 133), which is exemplified as follows: 12) Mary washed and dried the dishes. 13) John left yesterday. (Harlig & Bordovi-Harlig, 1988, p. 133) Left-dislocated referents are in a sentence-initial position; nevertheless, most of the left-dislocated referents in our data have stress. On the other hand, most of the pronominal or nominal references in the subject position in our data do not have stress. The fact might show the left-dislocated referents pragmatically have the marked stress. From this viewpoint, one may say that the speaker achieves their goal in the discourse, which the goal means that s/he topicalizes the shared information or emphasizes the thematic information, by not only the marked syntactic order but also the marked stress. Moreover, we can see that the speaker calls the hearer's attention to the thematic information by the stress. The left-dislocation causes a surprise for the hearer in the discourse context. In conclusion, this research on the discourse functions based on the referential distance and the topic persistence of the left-dislocations, and the discoursepragmatic functions based on the speaker's communicative intention in the usages of the leftdislocations, then we have descriptively seen that the applications of the left-dislocations could not be generalized down to one function. The fact agrees with the function of left-dislocation as the speaker's communicative strategy considered by Keenan-Ochs and Schieffelin (1976). However, our analysis and discussion about the discourse function as identifiability of the left-dislocated referents have more specified Keenan-Ochs and Schieffelin's study. # VI. # Conclusion The present study has attempted to descriptively examine the discourse-pragmatic functions of left-dislocation in American English discourse. The database for this study has consisted of three corpora of the Santa Barbara Corpus of Spoken American English, which has consisted of 77 minutes and 1753 clauses. The total percentage of left-dislocation out of the total number of clauses was 1% of the corpus. It was not high against our expectations. This paper analyzed the referential distance in the preceding discourse context and the topic persistence in the succeeding discourse context of the left-dislocations. Then, it divided the discourse functions of the left-dislocations into four types. We have clarified all left-dislocated referents in our data were identifiable for the hearer in the discourse context. Next, we have noted that the four discourse functions of the left-dislocations we analyzed were generalized down to the two different discoursepragmatic functions. We have suggested that the speaker tried to achieve their goal in the interactional discourse by using each discourse-pragmatic application of left-dislocation. To conclude, this study claims how marked syntactic structures interact with a particular discourse function, in addition, with a speaker's communicative goal in discourse context. Further studies are needed to examine this function and interaction-oriented grammatical aspects. In the future, we would like to investigate operations of left-dislocations in a larger database, and investigate other grammatical devices such as topicalization or thematization in discourse. 16. .. Each one was fighting for our affection [s].17. LENORE: [%]18. ALINA: ? That was very apparent.19. I knew that was gonna happen.20. (H) Plus they didn't like each other.21. (H) !Spar---22. Well !Spargo was ol=der,23. ? you know,4. .. Period.5. (H) And !Spargo was okay.6. .. (H) And we go out,7. ? and,8. so, I walk in,9. .. I see two paddlers... and these guys---,they were at each other's throats,12. 14. LENORE: ? Really.15. ALINA: .. Oh, yeah. 1Type30110144% 30 <70100%50%22% 20 <2533% 10<15Type1287229% 16 <3029% 11<1535%71% 5<1571% 5<10Type1113100%100%5%Type1124100%100%10%Total20 2Type231010155% 16 <2533% 11<1550%44% 10 <1566% 5<10Type007235%Type4213100%100%5%Type--2410%Total20 3Type 1Proper Noun910Noun with150%Genitive modifierType 2Noun with57Demonstrative35%'that'Noun with2Definite article'the'Type 3Noun with11Demonstrative5%'that'Type 4Indefinite22Noun10%Total20 4StrongWeakType 1821080%20%50%Type 261786%14%35%Type 3011100%5%Type 4022100%10%Total20 5% 6% * ALINA: remember !Tyke? * Lived next door to Mom? * Lenore: * % ? Yeah= ALINA: ? Okay * Two weeks ago I'm watching TV, 6 (h) and David Horowitz is going to have, 7. this former car .. radio thief on? 8. LENORE: ? It's her boyfriend * ALINA: (H) .. Yeah, 10. her ex-boyfriend * ALINA: (H) Well, 15. ? cause well, 16. .. he -17. .. he was a cocaine addict. 18. So he's talking about, 19. he's, 21. yeah, man, 22. he's gonna show us ! Mike 13. LENORE: ? How do you know? 14 23. .. you know, 24. how X not X, 25. LENORE: (THROAT * ALINA: to protect your car 27 not to get it, 28. you know, 29. ripped off man * I -33. .. yeah, 34. I was into it, 35. uh let me show you how easy it was VOX>. 36. (H) He's actually pretty intelligent 32 Cause, 31. you know * You know he just * ? (H) uneducated, 39. so, 40. .. u=m, 41. .. (TSK) yeah=. 42. ~Hector's radio=, 43. it was bro=ken, 44. we were gonna s - * ALINA: %Th-.. the friend that was there with them, 2. is this older guy with this young chick * LENORE: (H) * ALINA: ? hair=] pulled back, 9 in little pony [2tail=2 * VOX>, 12. (H) he said, 13. I would have been here, 14. but , 16. she's going, 17. (H) * Nothing was gonna help her. 23. (H) No make up, 24. no nothing, 25. Cause she's the little * you know, some, 33. .. (H) Lenore: 35. ALINA: [(DRINK) * LENORE: [(H) She's probably like] twenty-six References Références Referencias * Givenness, contrastiveness, definiteness, subjects, topics and point of view WallaceChafe Subject and topic NCharles Li NY Academic Press 1976 * Cognitive constraints on information flow WallaceChafe Coherence and Grounding in discourse RussellTomlin Amsterdam John Benjamins 1987 * Discourse, consciousness, and time: the flow and displacement of conscious experience in speaking and writing WallaceChafe 1994 University of Chicago Press Chicago * Representing hierarchy: constituent structure for discourse databases DuBois JohnWStephan Schuetze-Coburn Talking data: transcription and coding in discourse research JEdwards MLampert Eds Hillsdale Erlbaum 1993 * An outline of discourse transcription DuBois JohnW SusannaCumming & Danae Paolino J. Edwards & M. Lampert. 1993 * Leftdislocation in Italian conversation AlessandroDuranti Elinor Ochs Syntax and semantics TalmyGivón NY Academic Press 1979 12 discourse and syntax * BarbaraAFox Discourse structure and anaphora: written and conversational English Cambridge Cambridge University Press 1993 * Topic continuity in discourse: an introduction TalmyGivón Topic continuity in discourse: a quantitative cross-language study TalmyGivón Amsterdam John Benjamins 1983 * The grammar of referential coherence as mental processing instructions. Linguistics TalmyGivón 1992 30 * Syntax: an introduction TalmyGivón 2001 John Benjamins 1 Amsterdam * Shared knowledge and topicality JeanetteGundel Journal of pragmatics 9 1985 * Cognitive status and the form of referring expressions in discourse JNancyGundel Hedberg & Ron Zacharski Language 69 2 1993 * Notes on transitivity and theme in English, part 2 MichaelA KHalliday Journal of linguistics 3 1967 * Michael&Halliday RHasan Cohesion in English 1976 Longman * Accentuation typology, word order and themerheme structure Jeffrey & KathleenHarlig Bardovi-Harlig Studies in syntactic typology Amsterdam John Benjamins 1988 * Foregrounding referents: a reconsideration of leftdislocation in discourse EB BKeenan-Ochs Schieffelin 1976 2 * Wh-clefts and left-dislocation in English conversation: cases of topicalization Kyu-HyunKim Word order in discourse PDowing & M. Nooman Eds : JohnAmsterdam Benjamins 1995 * Information structure and sentence form KnudLambrecht 1994 Cambridge University Press Cambridge * Unattached NPs in English conversation Ono ASandra Thompson BLS 20 1994 * Toward a taxonomy of given-new information EllenFPrince Radical pragmatics Peter Cole NY Academic Press 1981 * Fancy syntax and shared knowledge EllenFPrince Journal of pragmatics 9 1985 * On the functions of leftdislocation in English discourse EllenFPrince Directions in functional linguistics AKamio Amsterdam: John Benjamins 1997 * Pragmatics and linguistics: an analysis of sentence topics TanyaReinhart Philosophica 27 1981 * Constrains on variables in syntax. The doctorical dissertation JRRoss 1967 at MIT * Goiron ShozoShibata 1975 Taishukan Publishing Company * Basic word order: functional principles RussellSTomlin 1986 Croom Helm London * Focal attention, voice, and word order: an experimental, cross-linguistic study RussellSTomlin Word order in discourse 1995