I win“ K VERSI SITY LIBRARI I58 IIIIIIIIIIIIIIIIII IIIIIIIII II II III IIII 3 1293 00909 This is to certify that the thesis entitled Topicalization in Japanese presented by Kelly Jean Fowler has been accepted towards fulfillment of the requirements for Master's . Linguistics degree in (at :56an fig; 45/ Major proésor/ Date //”/§' ?1 0-7639 MS U is an Affirmative Action/Equal Opportunity Institution r ‘N LIBRARY Michigan State University '\ ’- PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. __I ___I _JI___ . __II____I __][___ fiI—fI = I MSU Is An Affirmative Action/Equal Opportunity Institution swans-m TOPICALIZATION IN JAPANESE By Kelly Jean Fowler A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Linguistics and Germanic, Slavic, Asian and African Languages 1991 ABSTRACT TOPICALIZATION IN JAPANESE By Kelly Jean Fowler Despite the large volume of literature which has been written on topicalization in Japanese, a certain amount of controversy still exists. This may be partly due to the fact that most of these studies focus on oneparticular aspect of topicalization, and so do not allow for generalizations. This paper attempts to more clearly define topicalization by using a broader point of view. First, the parameter of subject/topic prominence is used as a framework in which to analyze Japanese topicalization, and evidence is given in support of the claim that Japanese is both subject and topic prominent. Next, this parameter is taken to the syntactic level, and it is suggested that topics are best analyzed as moving to COMP. Finally, this analysis is further supported by looking at the way case is assigned to topic noun phrases. TABLE OF CONTENTS Introduction I. Usage A. The Topic Marker Wa B. The Subject Marker Ga ll. Subject and Topic A. Background B. Distinguishing Subjects and Topics C. Characteristics of Toic-Prominent Languages D. The Basic Sentence Ill. Syntactic Structure of Topic Constructions A. The Chinese Focus Marker B. The Japanese Topic Marker IV. Topic Constructions and Case A. Kuno's Analysis B. Farmer's Analysis C. Kitagawa‘s Analysis Conclusion Bibliography iii 12 14 14 18 27 33 38 39 46 51 51 53 56 59 61 INTRODUCTION The role of topic constructions in Japanese has been an integral part of a variety of linguistic studies. On a theoretical level, Kuroda (1976, 1979) has discussed the traditional concept of the subject in connection to Japanese topics. At the descriptive level, reference grammars of Japanese have examined the usage of the topic marker wa, particularly in relation to the subject marker ga (Kuno (1973) and Yamagiwa (1942)). We has also been an integral part of discussions in other areas of Japanese grammar such as relativization (McCawley (1976), Kuno (1973) and Kitagawa (1982)), complementation (Josephs (1976) and Nakau (1973)), and subject-raising (Kuno (1976) and Nakau (1973)). Finally, Japanese topic constructions have been examined in relation to case (Kitagawa (1982) Farmer (1984), and Ostler (1980)). Despite the fact that topicalization in Japanese has stimulated a considerable number of studies over the past several decades, a certain amount of controversy still exists. This may be partly due to the fact that most of these studies have focused on a particular aspect of topicalization, and so do not allow for generalizations. In this regard, a broader linguistic point of view may be a useful approach. As a rule, generality and simplicity are desirable components of any scientific theory because they increase its explanatory and predictive powers (Darden, 1991, p. 413). Regarding linguistics specifically, Sells states, It is a goal of syntactic theory to provide a descriptive space within which the range of variation that we find among languages is precisely captured. That is, we would like to have a theory which is flexible enough to allow us to characterize all the fine variation we find, while still not allowing us to even consider certain possibilities (1985, p. 5). As an example of a possibility that syntactic theory should not allow for, Sells cites the fact that there is no natural language "in which questions are formed from normal sentences by reversing all the words in the sentence” (lbid). Thus, languages do not exhibit unlimited variation, but fall within definable scopes, or ”parameters" (lbid, p. 26). Furthermore, an attempt to define linguistic parameters is a useful scientific endeavor because it allows a given theory to abstract away from language-specific details and make broader hypotheses about languages in general. Using a generative framework, this paper will show that the typological parameter of subject/topic prominence can serve as a 6 useful framework for analyzing topic constructions in Japanese. By way of introduction, section I will describe possible usages of the topic marker wa, particularly in relation to the subject marker ga. Section II will provide evidence why Japanese should be classified as both subject-prominent and topic prominent. Section III will take this parameter to the syntactic level, and suggest a possible analysis for topic constructions. This syntactic analysis will be further supported in section IV by looking at the way case is assigned to topic noun phrases. l. USAGE Except for the fact that verbs are generally sentence-final, the basic word order in Japanese is fairly free (Kuno, 1973, p. 3). Grammatical relations are expressed by particles which follow the noun phrase (particles also have other functions which will not be discussed in this paper). The discussion in this section is intended to be descriptive rather than explanatory. A. The Topic Marker Wa It is not easy to define the usage of topics. Back in 1966, Pei defined the topic as ”roughly, for most languages, what traditional grammar defines as the subject" (p. 279). This vague definition may have been motivated by the fact that in Indo-European languages, the topic and the subject are often identical (Dubois, 1973, p. 490). However, according to Dubois, "on appelle topique le sujet du discours” (lbid). Crystal (1985) also defines topic at the discourse level; according to him, "the usefulness of the (topic/comment) distinction is that it enables general statements to be made about relationships between sentences which the subject/predicate 8 distinction . . . obscures" (p. 311). In other words, while the role of the subject can be defined in relation to the sentence in which it is contained, in order to define the role of the topic, it may be necessary to refer back to previous sentences in the discourse. The discussion which follows should clarify this. In Japanese, the particle we can have two functions: topicalization or contrastiveness (Kuno (1973) and Kitagawa (1982»: (1) John wa sono hon o yonda. TOP that book OBJ read 'As for John, he read that book.‘ (2) John wa sono hon o yonda ga Mary wa yomanakatta. CONT that book OBJ read but CONT didn't read 'John read that book but Mary didn't.‘ (Kuno, 1973, p. 47) In (1), we topicalizes the NP 'John,‘ whereas in (2), the we particles set up a contrast between John and Mary. It should be observed that in (1), the topic appears to be synonymous with the subject. However, this is not always the case: (3) siizaa wa buruutasu ga korosita. Caesar TOP Brutus SUBJ killed 'As for Caesar, Brutus killed (him).' (Nakau, 1973, p. 49) 9 In (3), 'Caesar' is the topic, and the subject position is filled by 'Brutus.‘ It should also be noted that contrastive wa does not necessarily require the presence of two overt wa markers: (4) Ame wa hutte imasu ga . . . Rain CONT falling is but 'lt's raining but . . . ' (Kuno,1973,p.38) How do we know that the we in (4) is contrastive and not topical? According to Kuno, ”the themes (topics) of Japanese sentences must be either generic or anaphoric. The contrastive we, on the other hand, can place non-anaphoric noun phrases in contrast" (lbid, p. 46). Sentence (4) is not generic because it is about specific rain, that which is falling now. Example (5) shows rain used as a generic topic: (5) Ame wa sora kara huru. Rain TOP sky from falls 'Speaking of rain, it falls from the sky.’ The question of whether or not (4) has an anaphoric topic is somewhat confusing because of the way Kuno uses the term 'anaphoric.’ In generative linguistics, an anaphor is "a type of noun phrase which has no independent reference, but refers to some other 1 0 sentence constituent . . . an anaphor must be bound in its governing category” (Crystal, 1985, p. 17). In other words, an anaphor and its antecedent must be contained in the same sentence. Kuno, however, obviously intends the term 'anaphor' to refer to an NP mentioned previously in the discourse. Throughout the rest of this paper, the term 'anaphor' will refer to Kuno's usage. Returning to (4), the topic of this sentence could or could not be anaphoric, depending on whether rain had been mentioned previously in the discourse. This means that the we in this sentence could be either contrastive or topical. Thus, every clause containing a we marker may not be clearly contrastive or topical. Sentence (6), for example, is also ambiguous. (6) Watakusi ga sitte iru hito wa paati ni kimasen deshita I SUBJ know people TOP/CONT party to didn't come (a) 'Speaking of the persons whom I know, they did not come to the party.‘ (b) (People came to the party, but) there were none I knew.‘ (Ibid, p. 48) The ambiguous phrase in (6) is watakusi ga sitte iru hito we, 'the people I know.‘ In interpretation (a), the phrase has anaphoric reference. (Again, it must be assumed that "the persons whom I 11 know" was mentioned previously in the discourse). Thus, wa marks the topic. In interpretation (b), the speaker is talking about people who have not previously been mentioned in the discourse. This reference is therefore nonanaphoric, and so we marks contrastiveness. A sentence may also have both a topic we and a contrastive we: (7) Bunmeikoku wadanseiwa keikin jumyoo ga nagai. civilized nations TOP man CONT average lifespan SUBJ long 'As for civilized countries, as far as men are concerned, their average lifespan is long.‘ (Kitagawa, 1988, p. 189) Finally, noun phrases are not the only category which can be topicalized. In (8), a postpositional phrase is the topic: (8) Amerika de wa Sumiko ga kuruma o katta America at TOP SUBJ car OBJ bought 'In America, Sumiko bought a car. (lbid, p. 191) Example (9) shows an adverb which has been topicalized: (9) Kinoo wa Taroo ga hirumesi o tsukutta Yesterday TOP Taro SUBJ lunch OBJ made 'Yesterday, Taro made lunch.‘ (lbid, p. 192) 12 B. The Subject Marker Ga The marker ga has three possible functions (Kuno (1973) and Kuroda (1979)). The first is neutral or “descriptive ga " in subject position: (10) Arne ga hutte iru rain SUBJ falling is 'lt's raining.‘ (Kuno, 1973, p. 50) Descriptive ga is limited to sentences with “action verbs, existential verbs, and adjective/nominal adjectives that represent changing states" (lbid, p. 49). The second usage of ya , ”exhaustive-listing ga," occurs with stative predicates: (11) John ga gakusei desu. SUBJ student is '(Of all the people we are talking about) John (and only John) is a student.‘ (lbid, p. 51) Finally, ga may function as an object marker: (12) John wa Mary ga suki desu. TOP OBJ fond of is 'Speaking of John, he likes Mary.’ (lbid, p. 55) It should be noted that the marker 0 may also mark an object (see sentences 1, 2, 8, 9). 13 Summary To summarize, we may mark either topicalization or contrastiveness, while ga may have one of three functions: descriptive or exhaustive listing in subject position, or object-marking. II. SUBJECT AND TOPIC A. Background Present-day linguistic theory recognizes the subject—topic dichotomy as a typological distinction of world languages (Rutherford, 1983). However, earlier linguistic studies were mainly concerned with the notion of subject. This is because ”the traditional concept of subject . . . originated in the scholarly tradition of speakers of lndo-European languages.“ (Kuroda, 1976, p. 10). In other words, the first studies were done on languages which were subject-prominent rather than topic-prominent. As a result, "the structure of the major languages of western civilization has led to an erroneous emphasis on the role of the subject in language” (Lehmann, 1976). Furthermore, the definition of a subject is also biased toward lndo-European languages. For example, according to Kuroda (1976, p. 10), traditional grammar (Port-Royal grammar) describes the surface subject as "that constituent of a sentence with which the main verb agrees in number and person.” Such a definition has no relevance for a language like Japanese, in which there is no subject-verb agreement. 14 15 However, according to Kuroda (1976, p. 5), 'the traditional notion of subject-predicate structure” has little relevance for most present-day logicians and linguists. In his opinion, though, Japanese shows two different kinds of sentences, one with a subject-predicate structure, and the other without. He gives the following examples: (13) lnu ga hashitteiru. dog NOM running is 'A/the dog(s) is/are running.‘ (14) lnu wa hashitteiru dog TOP running is 'The dog(s) is/are running.’ (Kuroda, 1976, p. 6) According to Kuroda, a sentence with an initial wa phrase expresses 'the subject of the judgment of the subject-predicate form,” while those like (13) which have no we phrase express a thetic judgment (lbid). This analysis is vague, since Kuroda never clearly defines the exact manner in which he is using terms such as ”judgment" and "thetic.” Furthermore, he himself admits that the empirical evidence is weak. After a one-page analysis regarding definiteness and specificity, he concludes: 16 But there is, admittedly, not much more that I can do to support the claim that a sentence with a sentence-initial wa phrase represents a judgment of the subject-predicate structure, while a sentence without a subject-initial wa phrase represents a judgment with the subject-predicate structure, except appealing directly to native intuition. (Kuroda, 1976). Kuroda appears to have captured the fact that there are two general sentences types in Japanese. However, his theory lacks a formalized distinction. A closer analysis of the subject-topic distinction, such as the one Li and Thompson provide in their 1976 paper, ”Subject and Topic: A New Typology of Language," may be able to provide such a formalization. Rather than just a dichotomy between subject-predicator and topic-comment, Li and Thompson suggest that languages of 'the world show a four-way classification: (1) subject prominent, (2) topic prominent, (3) both subject-prominent and topic-prominent (both subjects and topics can be distinguished) and (4) neither subject-prominent nor topic-prominent (subjects and topics cannot be distinguished from each other) (Li and Thompson, 1976, p. 459). According to them, ”a synchronic typology is shown in which different languages are caught at various stages." (lbid). In other words, languages are continually evolving from one stage to another. Support of this theory is Lehmann's analysis of the 17 historical development of lndo-European languages: If there were a syntactic change, leading to the requirement that a grammatical subject be expressed in a matrix sentence, the item often expressed as topic would be the subject. A topic-prominent language would in this way develop into a subject-prominent language. . . I propose that just such a develoment took place in lndo-European. (Lehmann, 1976, p. 450). Lehmann then goes on to outline the development, using Li and Thompson's criteria as a guideline. Thus, the evidence for the subject-topic distinction as a continuum rather than a dichotomy is supported historically. Without this continuum, it would be difficult to support the validity of language types (3) and (4). This is important because according to this analysis, Japanese should be considered language-type (3), both subject-prominent and topic-prominent. Japanese will now be examined according to the points which Li and Thompson offer for distinguishing subjects and topics. 1 8 B. Distinguishing Subjects and Topics 1. Definiteness Topics, including both proper and generic noun phrases, must be definite, while subjects may or may not be (Li and Thompson, 1976). Definiteness may be defined as "referents which are identifiable not only by their name but by a description which is sufficiently detailed to enable that referent to be distinguished from all others, e.g. the present Queen of England . . . Definiteness in English is usually conveyed through the use of definite DETERMINERS (such as this, my), and especially through the definite article the" (Crystal, 1985, p. 86). Kuno disagrees with Li and Thompson on this point. He states, ”what determines whether a specific noun phrase can become a topic or not is whether the noun phrase is anaphoric and not whether it is definite" (1976, p. 40). He gives the following English example: (15) I know the man who killed Robert Kennedy (lbid, p. 41). According to Kuno, in this sentence, the does not make the noun phrase anaphoric; it merely indicates that only one man killed 1 9 Kennedy. Replacing the with a would indicate that a group of men had killed Kennedy. As evidence that this sentence is not‘anaphoric, Kuno cites the fact that this noun phrase could not be topicalized: (16) #Speaking of the man who killed Robert Kennedy, he does not seem to have been involved in any conspiracy (lbid). This sentence would be ill-formed if the listener didn't know that Robert Kennedy had been killed. Since this topic marker meets the definiteness criteria, however, Kuno attributes the problem to the fact that it is not anaphoric. Along these same lines, Chafe (1976) uses the term ”givenness." According to him, ”given (or old) information is that knowledge which the speaker assumes to be in the consciousness of the addressee at the time of the utterance” (p. 30). He cites the Japanese markers we and ya as the best known examples of the given-new distinction. Regarding definiteness Chafe states, ”there is no reason they (definite items) cannot be either given or new . . . although definiteness and givenness often go together” (lbid). Thus, it appears that both Kuno and Chafe agree that, while topics in Japanese are generally definite, this does not cover all cases. Li and Thompson's point may be more accurately illustrated in Japanese with another term such as anaphoric or given. 20 2. Selectional Relations A topic need not be an argument of a verb, but a subject will ”always have a selectional relation with some predicate in the sentence.” (Li and Thompson, 1976, p. 460). This is in line with many analyses of case-marking in Japanese, including Kitagawa (1982), and Farmer (1984). According to Kitagawa, ”to each argument in the PAS (Predicate Argument Structure), the 'case linking rules' assign linking registers signified by ’case particles‘ (GA, 0, NI)” (p. 183). Thus, the subject marker ga is part of the predicate argument structure. As for topics, Kitagawa states, ”we indicates that the immediately preceding X' is outside the domain of 'evaluation' in terms of the PAS of the nucleus V" (p. 184). In other words, the noun phrase which precedes wa will not be assigned an argument role by the verb. This situation is shown in (17): 17) Taroo wa Hanako ga iedeshita TOP SUBJ leave home-do-past 'As for Taro, Hanako ran away from home.’ (Farmer, 1984, p. 86) This sentence has only one argument slot in the PAS, and this is taken by the subject NP, Hanako. There is no argument slot for the topic NP. The problem of assigning case to the topic marker will be discussed further in section N. For now it should suffice to note 21 that, as Li and Thompson suggest, a topic need not be part of the predicate argument structure of the verb. 3. Subcategorization of Verbs In this criterion, Li and Thompson are concerned with the fact that specific verbs will subcategorize for subjects with specific thematic roles. The topic, on the other hand, though it may be determined by discourse, is definitely not determined by the verb. (Li and Thompson, 1976). The following example illustrates this point in Japanese: (18) Nihon - ni - wa kankoo-kyaku ga oozei kuru Japan to TOP tourists SUBJ many come 'To Japan, many tourists come.‘ (Kuroda, 1979, p. 57) The subject phrase kankoo-kyaku ga is categorized for by the verb, and could not be replaced, for example, by an instrumental NP such as naifu de, 'by a knife,‘ or an object NP such as udon o, 'noodles.’ On the other hand, the tapic NP is freer; it could be replaced by a temporal topic such as haru ni wa, 'in the spring,’ an instrumental NP such as basu de, 'by bus,‘ or it could be omitted completely. Thus, it can be concluded that the distinction Li and Thompson are 22 making here regarding subcategorization for subjects is clear in Japanese. 4. Functional Role According to Li and Thompson, "the functional role of topics is constant across sentences” in that it sets up a domain for the predicate (Li and Thompson, 1976, p. 463). This is similar to Chafe who states that “what the topic appears to do is to limit the applicability of the main predication to a certain restricted domain.” (Chafe, 1976, p. 50). The following are examples of this in Japanese: (19) Sakana wa tai ga ii fish TOP red snapper SUBJ good 'Speaking of fish, red snapper is the best.‘ (Kuno, 1973, p. 50) (20) Hana wa sakura ga ii flower TOP cherry blossom SUBJ good 'Speaking of flowers, cherry blossoms are the best.‘ (lbid, p. 51) In (19), the predicate, that red snapper is the best, is within the domain of fish. Similarly, in (20), the predicate that cherry blossoms are the best is within the domain of flowers. As for subject NP's, they may or may not play a semantic role. Furthermore, 23 The functional role of the subject can be defined within the confines of a sentence as opposed to discourse. According to Michael Noonan, the subject can be characterized as providing the orientation or the point of view of the action, experience, state, etc., denoted by the verb. The difference in the functional roles between the subject and the topic explains the fact that the subject is always an argument of the verb, while the topic need not be. (Li and Thompson, 1976, p. 464). Li and Thompson cite empty and dummy subjects as examples of subjects without a semantic role. However, these phenomena do not occur in Japanese. According to Li and Thompson, this is typical in topic-prominent languages because 'the subject does not play such a prominent role.” (lbid). The fact that the functional role of the subject can be defined within the domain of the sentence and is thus always related to an argument was discussed under part two, selectional relations. 5. Verb Agreement A language may have obligatory agreement between the subject and the verb. However, ”agreement between the topic and the predicate is very rare.” (Li and Thompson, 1976, p. 464). This does 24 not apply to Japanese, which does not show morophological or syntactical subject-verb agreement. 6. Sentence-Initial Position In all the languages examined by Li and Thompson, topics were always found in sentence-initial position. In contrast, subjects could be but were not necessarily in sentence-initial position. It is true that in Japanese the topic is generally sentence-initial: (21) boku wa kono giron ga itchiban settokuteki da to omou I TOP this argument SUBJ most persuasive COP CM believe 'I believe this argument is the most persuasive.‘ (Kuroda, 1988, p. 133) (Note: COP .. copula CM - complement marker) Here, the topic NP is sentence-initial, in line with Li and Thompson. However, Kuroda notes that the following sentence is also acceptable: (22) kono giron ga boku wa itchiban settokuteki da to omou this argument SUBJ l TOP most persuasive COP CM believe 'I believe this argument is the most persuasive.’ (lbid) In this sentence, boku wa, the topic NP, has been moved from 25 sentence-initial position into the subordinate clause. Kuroda's analysis is that ”the topic has been downgraded, that is, moved into the subordinate clause from the clause-initial position" (lbid, p. 133). Another possible analysis is that the subject has been proposed to the front of the sentence. However, the important point here is that sentences (21) and (22) have the same deep structure, but that in the surface structure of (22), movement has taken place. Thus, regarding the characterization that topics always occur sentence-initially, it might be more accurate to say that in Japanese, topics occur sentence-initially in deep structure. As for subjects, their position in a Japanese sentence is somewhat freer. If a sentence has no topic then the subject may occur first, but this is not obligatory: (23) Taroo ga Hanako ni sono hon o ageta SUBJ to that book OBJ gave 'Taro gave that book to Hanako.‘ (24) Taroo ga sono hon o Hanako ni ageta. (25) Sono hon o Taroo ga Hanako ni ageta. (Farmer, 1984, p. 73) Sentence (23) contains three NP's and a verb in sentence-final position. Sentences (24) and (25) show that changing the order not only of the subject NP, but of any of the NP's does not alter the 26 grammaticality or the meaning of the sentence, although there is a shift in focus. 7. Grammatical Processes Li and Thompson state that, ”the subject but not the topic plays a role in such processes as reflexivization, passivization, Equi-NP deletion, verb serialization, and imperativization (Li and Thompson, 1976, p. 465). In order to illustrate this point, reflexivization will be taken as an example. According to Kuno, in English, while several elements may serve as the antecedent of a reflexive pronoun, in Japanese only the subject may be the antecedent of a reflexive pronoun: (26) Bill ni wa John ga jibun no shashin o miseta to TOP SUBJ self POSS pictures OBJ showed 'To Bill, John showed pictures of himself.‘ In the Japanese form of this sentence, the reflexive pronoun can only refer to the subject, John, and not to Bill, which is in topic position. Furthermore, placing the topic between the subject and the reflexive pronoun does not alter the interpretation: 27 (27) John ga Bill ni wa jibun no shashin o miseta ' SUBJ to TOP himself POSS pictures OJB showed 'John showed Bill pictures of himself.‘ In contrast, in the English equivalent of (27), the reflexive pronoun could refer to either John or Bill. This supports Li and Thompson. Summary To summarize this section, Li and Thompon's seven points are useful for distinguishing subject and topic in Japanese. Definiteness, selectional restrictions, thematic roles, the functional role, the sentence-initial position and grammatical roles all support the theory that Japanese is indeed topic-prominent. Only one point, subject-verb agreement is not relevant to Japanese. C. Characteristics of Topic-Prominent Languages Li and Thompson also list several aspects of topic-prominent languages in general. The following is an analysis of these points as regards their relevance to Japanese: 28 1. Surface Coding First, Li and Thompson note that topic-prominent languages code for topic but not for subject. It is interesting to note that Japanese codes both for topic (wa) and subject (ga). This may be argued to be evidence for the claim that Japanese is both tapic-prominent and subject-prominent. 2. Passive Constructions Passive constructions are less common in topic-prominent languages than in subject-prominent languages because the subject is so crucial to this construction: The notion of-subject is such a basic one that if a noun other than the one which a given verb designates as its subject becomes the subject, the verb must be marked to signal this 'non-normal' subject choice. (Li and Thompson, 1976, p. 467). According to Li and Thompson, in Japanese, the passive does exist but carries the special meaning of adversity (lbid). Kuno distinguishes two passive forms in Japanese: pure and adversity. Adversity passives are defined as "sentences that imply that the 29 subject of the main sentence has been inadvertently affected by the action represented in the rest of the sentence” (Kuno, 1973, p. 302). Sentence (28) is an example of a pure passive, and sentence (29) of a adversity passive: (28) Sono yubiwa wa doroboo ni nusurareta that ring TOP thief by was stolen 'The ring was stolen by a thief.‘ (29) Mary wa sono yubiwa o doroboo ni nusurareta TOP that ring OBJ thief by stolen 'Mary had the ring stolen on her by a thief.‘ (lbid, p. 302) The adversity passive "arises only when there is an extra noun phrase which cannot be accounted for by the simplex deep structure of the active version." (lbid). In (29), the extra NP is Mary wa. Kuno's account for these two types of passive is that they have different deep structures. According to his analysis, the 08 of (28) and (29) respectively would be as follows: (32) doroboo ga sono yubiwa o nusunda thief TOP that ring OBJ stole (33) Mary wa[doroboo ga yubiwa o nusunda] TOP thief SUBJ ring OBJ stole Kuno uses arguments concerning possible insertion and interpretation of reflexive pronouns to support this analysis. What 30 is interesting to note for the purposes of this paper is that there appear to be two distinct passive patterns in Japanese: one which is used for sentences with topics, and one for sentences without. 3. Dummy Subjects The third characteristic of topic-prominent languages is that they seldom invoke dummy subjects such as it in 'It is raining.’ This is because in a subject-prominent language a subject may be needed whether or not it plays a semantic role . . . In a topic-prominent language, as we emphasized, where the notion of subject does not play a prominent role, there is no need for ”dummy subjects." (Li and Thompson, 1976, p. 467) Since Japanese does not employ dummy subjects, as was discussed in part four of section one, it is in accordance with this characteristic. 4. Double Subjects A sentence with both a topic and a subject is termed a ”double subject” construction by Li and Thompson. Sentences of this kind include two already mentioned in this paper: (19) and (20). The 31 following is a similar example from Li and Thompson: (32) Sakana wa tai ga oishii fish TOP red snapper SUBJ delicious 'As for fish, red snapper is delicious.’ (Li and Thompson, 1976, p. 468) This kind of sentence, which clearly shows both a topic and a subject, is evidence that "no argument can be given that these sentences could be derived by any kind of 'movement' rule from some other sentence type" (lbid). In other words, the co-occurrence of a topic and a subject indicates that the topic could not be in, for example, subject position at deep structure, since this position is already filled. This idea will be elaborated on later in the section on basic sentences. 5. Controlling Coreference Using the following example from Mandarin, Li and Thompson state that in a topic-prominent language, ”the topic, and not the subject, typically controls co-referential constituent deletionz' (33) Nei kuai tian daozi zhangde hen da, suoyi hen zhiqian that piece land rice grow very big so very valuable 'As for that piece of land, rice grows very big, so it (the land) is very valuable.‘ (p. 469). 32 The missing subject in the second clause can only refer to the topic, 'land,’ and not to the subject, 'rice.' A similar case can be made for Japanese: (34) Sono tochi wa kome ga yoku sodatsu node «totemo kachi that land TOP rice SUBJ well grows so very valuable ga aru SUBJ COP 'As for that piece of land, rice grows very big, so it (the land) is very valuable.‘ (p. 469). Sentence (34) has the same meaning as the Chinese equivalent (33) in that the empty constituent must refer to the topic, not the subject. In English, a subject-prominent language, there are two possible interpretations. Thus, once again, this evidence argues that Japanese is a tapic-prominent language. 6. Verchinal Languages Topic-prominent languages are generally verb-final. The basic unmarked word order of Japanese is also verb-final, as evidenced by any of the sentences above, but particularly in examples (23) to (25), where the free order of the noun phrases sounds out in contrast to the more rigid positioning of the verb. 33 Summary In conclusion, the above discussion regarding characteristics of topic-prominent languages supports the analysis of Japanese as topic-prominent in the areas of surface coding, passive constructions, double subjects, controlling co-reference, and v-final languages. In addition, the areas of surface coding, passive constructions and double subjects provide dual support for the fact that Japanese is also a subject-prominent language. This is in accordance with the original postulations that Japanese is both subject-prominent and topic-prominent. D. The Basic Sentence Finally, the notion of the basic sentence can contribute support to the t0pic-prominent/subject-prominent issue. The question regarding Japanese is whether a sentence with a topic is derived from a sentence without a topic, or whether these are different kinds of sentences. For example, do (1) and (35) have 34 the same deep structure, or not? (1) John wa sono hon o yonda TOP that book OBJ read 'As for John, he read that book.’ (35) John ga sono hon o yonda SUBJ that book OBJ read 'John (and only John) read that book.‘ In these two sentences, the NP's containing 'John' have different grammatical functions. In (1), John has anaphoric reference, whereas in (35) ga ”exhaustively lists" John (see section I). According to Yamagiwa (1942), a topic may be regarded as the subject of a sentence if there is no NP-ga. This analysis implies that the two kinds of sentences are the same, 'and that NP movement takes place in sentences involving t0pic constructions. Although in (1) this would seem to be an acceptable analysis, since the topic and subject appear to be synonymous, as was pointed out in section 1, this is not always the case. Consider the following: (36) Kudamono wa Mary ga ringo o tabeta. fruit TOP SUBJ apple OBJ ate 'As for fruit, Mary ate an apple.’ Omitting the subject here still produces a grammatical sentence: (37) Kudamono wa ringo o tabeta fruit TOP apple OBJ ate 'As for fruit, (she) ate an apple.’ 35 However, to regard the topic as the subject in (37) would change the meaning of the sentence to something like, ”The fruit ate an apple.” Since this is clearly not the intended meaning, Yamigawa's analysis is unacceptable. There is more evidence to support the fact that sentences with topics are not simply derived from sentences without topics. For example, it has been shown that the passive sentence analysis requires that topic and nontopic sentences have different deep structures. Thus, it is reasonable to assume that sentences such as (1) be given a deep structure in which the subject category is either allowed to remain empty, or is omitted. According to part B-4 of this section, this is common in topic-prominent languages. Furthermore, this is in line with the general grammar of Japanese, which allows subject dropping or omission: (38) wakarimasen don't understand '(I) don't understand. (39) doko iku no where go OM 'Where are (you) going?‘ ( Note: OM - question marker) 36 Summary It seems clear from the above discussion that topics and subjects are different categories in Japanese. Although the NP-ga phrase seems to be linked to the predicate, the domain of the tapic appears to have a broader domain. Therefore the following analysis seems reasonable: 1) Noun phrases which are part of the predicate argument structure are sisters of V (such a structure has been proposed by many others in the past, including Farmer (1984): S' S NP—ga NP-o NP-ni V 2) Noun phrases with the topic marker we are sisters of S (This is in line with many analyses such as Kitagawa (1982), who claims that the topic NP is outside of the PAS: 37 SI S COMP NP-wa NP-ga VP Section III will defend the analysis that topic constructions belong in COMP position, and section IV will consider how to assign case to an NP in COMP position. III. SYNTACTIC STRUCTURE OF TOPIC CONSTRUCTIONS And the end of section II, it was proposed that topic noun phrases in Japanese move to COMP position. Since movement processes in generative grammar have traditionally been concerned with configurational languages such as English, such processes would seem to have little relevance for nonconfigurational languages like Japanese. Nevertheless, several recent studies have attempted to apply an abstract concept of movement to various components of Japanese grammar. One example is Saito and Hoji, who argue that postulating WH movement in Logical Form (LF) can account for weak crossover in Japanese (Saito and Hoji, 1984). As regards NP movement in Japanese, Farmer states, ”to say that English has NP-movement is simply to speak metaphorically. The question now is not 'Why doesn't Japanese have NP-movement?‘ but rather 'Can we extend the metaphor to Japanese, and if so, what are we picking out?" (Farmer, 1984, p. 196). Farmer's question is a crucial one. If applying metaphorical movement to Japanese is merely an attempt to squeeze a nonconfigurational language into the generative framework, then the metaphor is not a useful one. However, this section will argue that movement provides a more general analysis 38 39 of topicalization. Another nonconfigurational language which is said to lack movement is Chinese. However, in an article entitled WH-Movement, Huang attempts to show that the concept of move-alpha is nevertheless relevant to Chinese in LF. Under this analysis, Huang is able to show that Chinese obeys UG constraints such as Subjacency (Huang, 1981). This paper will look at some aspects of Japanese grammar in the light of Huang's analysis, focusing on how such a comparison can help define the role of U6 constraints in Japanese. In particular, the Chinese FOCUS marker shi will be compared to the Japanese TOPIC marker wa. In part (a), the relevant points of Huang's analysis will be summarized. Part (b) will compare the role of the Japanese topic marker to Chinese. A. The Chinese Focus Marker 1. Function The purpose of the FOCUS marker (FM) is to highlight a given construction in the clause. The following examples are from Huang: 40 (40) shi wo mingtian yao mai neiben shu FM l tomorrow want buy that book 'It is I that want to buy that book tomorrow.’ (41) wo shi mingtian yao mai neiben shu l FM tomorrow want buy that book 'It is tomorrow that I want to buy that book' (42) wo mingtian shi yao mai neiben shu I tomorrow FM want buy that book 'I do want to buy that book tomorrow.‘ (p. 372) The FOCUS construction does not involve the movement of any constituents. The FM simply precedes the constituent to be emphasized: we (I) in (40), mingtian (tomorrow) in (41), and yao (want) in (42). In contrast, the English equivalent is a cleft sentence in which the emphasized construction has been fronted. Thus, the FOCUS construction may be translated as 'It is l' in sentence one and 'It is tomorrow' in sentence two. In sentence three, a FOCUS verb is translated as 'I do want' because verbs cannot be clefted in English. According to Huang, the Chinese FOCUS construction is best analyzed as a cleft sentence in LF because, Since a cleft sentence has the universal semantic property of dichotomizing a sentence into focus 41 and presupposition, it is natural to provide a unified representation of this dichotomy in LF (p. 372). Sentence (40) thus has the following structure in LE: (43) [ ( shi wo ) x [ X mingtian yao mai neiben shu ] ] PM I tomorrow want buy that book 'It is I who want to buy that book tomorrow.‘ In this example, the FOCUS element has moved to the clause-initial COMP position and may be described as ”a quasi-quantifier binding a variable in a presupposition” (lbid). In other words, the sentence now has the structure of a cleft sentence in LF. In some situations, FOCUS can occur in an embedded clause: (44) Zhangsan shuo [ Lisi shi mingtian lai] say FM tomorrow come (a) 'Zhangsan said that it is tomorrow that Lisi will come.’ or (b) 'It is tomorrow that Zhangsan said that Lisi will come.‘ (p. 373) According to Huang, the reason this sentence has two interpretations is because there are two possible positions which the FM can move to. First, it can move to the COMP position of the embedded clause in which case it will have the meaning of 42 interpretation (a). Second, it can move to the COMP position of the entire sentence, in which case it will have the meaning of interpretation (b). 2. Constraints Based on the above analysis, the following constraints may be observed: Long distance clefting is possible from verb phrase complement positions, but not from a syntactic island like a complex NP or a sentential subject. This suggests that cleft formation in Chinese, although it does not involve any overt movement rule, nevertheless has to obey Subjacency (p. 375). Sentence (44) shows that FOCUS may occur in an embedded clause as the complement of a VP. This is not true of all VP complements, however. "The embedded clause must be, in some sense, a direct discourse complement to a verb or a noun . . . verbs that subcategorize for such a complement include shuo 'say,’ xiang 'think,’ etc.” (p. 409). Sentence (45) is an example of a violation of the Complex NP Constraint (CNPC), and sentence (46) is an example of a violation of 43 the Sentential Subject Constraint (SSC): (45) * [ wo xihuan [ shi Zhangsan mai de neizhi gou ] ] I like FM buy DE that dog * ' I like the dog that it is Zhangsan that bought.‘ (46) * [[Zhangsan shi mingtain lai]mei guanxi FM tomorrow come no matter *‘That it is tomorrow that Zhangsan will come does not matter.‘ (Note: DE is a relative clause marker) (p. 374) In sentence (45), FOCUS is contained in the complex NP shi Zhangsan mai de neizhi gou, 'the dog that it is Zhangsan that bought.‘ However, according to Huang's analysis, in LF, the FOCUS must move to the COMP position at the beginning of the clause. This violates the CNPC. Similarly, in sentence 46, FOCUS is contained in the sentential subject, Zhangsan shi mingtain lai, 'that it is tomorrow that Zhangsan will come,‘ and moving it to COMP would violate the SSC. This explains why both these sentences are ungrammatical. A sentence containing two FOCUS constructions is also ungrammatical: (47) * [ s [s shi Zhangsan shi mingtian yao lai ] ] FM FM tomorrow want come * 'It is Zhangsan that it is tomorrow that will come.‘ (p. 375) According to Huang, the ungrammaticality of this sentence can be 44 attributed to the fact that there are two FOCUS constructions in COMP position, creating a branching node. This means that "neither operator c-commands its variable” (p. 376). This example shows that, Application of FOCUS may be blocked not only by a syntactic island like a complex NP or a sentential subject, etc., but also by what may be called a 'Focus lsland' formed by a previous application of the same rule. (p. 377) The 'Focus lsland' then, is the third constraint blocking FOCUS movement. Finally, FOCUS movement may be blocked by a WH-lsland: (48) * [ s [ shi Zhangsan da-le shei ] ] ? FM beat who * 'Who is it Zhangsan that beat?‘ (p. 377) In order to explain the ungrammaticality of this sentence, it must also be assumed that there is WH-movement in Chinese at LF. For example, the surface structure of this sentence: (49) ni xihuan shei? you like who 'Who do you like?‘ would move the Wh-word to COMP position in LF: 45 (50) [ sheii [ ni xihuan 9i ] ] who you like (p. 370) Going back to sentence (48), the grammaticality of this construction can now be explained by the fact that COMP is a branching node containing shei (who) and FOCUS. Like the 'Focus Island,‘ this sentence is ungrammatical because neither the PM or the WH-word c-command their variables. In summary, Huang has argued that FOCUS should be analyzed as moving to COMP position in LF. This is in accordance with universal semantic principles of the cleft sentence. Furthermore, according to Huang's analysis, Chinese can be shewn to obey Subjacency. FOCUS cannot occur in a complex NP or a sentential subject because to do so would violate Subjacency. In addition FOCUS cannot occur in a clause with another FOCUS or in a WH-lsland because to do so would create a branching COMP node, which would violate c-commanding requirements. 46 B. The Japanese Topic Marker Like the Chinese FOCUS, TOPIC may sometimes occur in an embedded clause as the complement of a V. Verbs that subcategorize for this in Japanese include imasu, 'say,‘ and omoimasu, 'think.‘ The following examples illustrate this: (51) Zhangsan wa ashita wa Lisi ga kuru to iimashita TOP tomorrow TOP NOM come QM said 'As for Zhangsan, he said that, as for tomorrow, Lisi will come.‘ (52) Zhangsan wa ashita wa Lisi ga kuru to omoimasu TOP tomorrow TOP NOM come OM think 'As for Zhangsan, he thinks that, as for tomorrow, Lisi will come. (Note: OM is a quotation marker) In these examples, the first topic noun phrase, Zhangsan wa, has scope over the entire sentence. In contrast, the second topic noun phrase, ashita wa, only has scope over the embedded clause. This means that (52), for example, could not be interpreted to mean, 'As for Zhangsan, and as for tomorrow, he thinks that Lisi will come.‘ These rules show parallel structure with the Chinese example in (44). 47 Regarding the CNPC and the SSC, these are generally considered to be inapplicable to Japanese (Kuno, 1983). Sentence (53) is a violation of the CNPC: (53) [ [ kowaigatte-ita ] s inu ga sinde shimatta] s kodomo was fond of dog NOM died ended-up-with child 'the child who the dog (he) was fond of died.’ (lbid, p. 239) According to Kuno, the original complex NP sono kodomo ga kawaigatte-iru inu, 'the dog that the child was fond of,‘ has been relativized, and kodomo, the child, deleted. According to the CNPC, this sentence should be ungrammatical. The fact that the sentence is well-formed supports the theory that the CNPC does not apply in Japanese. It should be noted that this situation is a little ambiguous; Kuno notes that native-speaker intuitions on these kinds of sentences vary (lbid, p. 240). Sentence (54) illustrates a violation of the SSC in Japanese: (54) [watakushi ga au koto] ga muzukashii hito l person to meet that NOM difficult 'the person whom that I see him is difficult' (lbid, p. 241) According to the SSC, (54) should be ungrammatical because muzukashii hito has been moved out of the original sentential subject. The fact that the sentence is well-formed indicates that 48 the SSC does not apply to Japanese. Despite this situation, complex NP's and sentential subjects are still relevant concepts in Japanese. If we is considered to be a long distance clefting structure like FOCUS in Chinese, then some interesting results can be found. Basically, we can never be found inside a complex NP or a sentential subject: (55) * kore wa [watakushi wa kaita ] hon desu this TOP l TOP wrote book COPULA *‘As for this, it's, as for me, the book that I wrote.‘ (56) kore wa [ watakushi ga kaita ] hon desu this TOP l - NOM wrote book COPULA 'As for this, it's the book that I wrote. (Kuno, 1983, p. 234) Another similarity between Chinese FOCUS and Japanese TOPIC is that both can only appear once in a clause. According to Kuno, "a given sentence can only have one thematic (topical) wa: if there is more than one occurrence of we in a sentence, only the first can be thematic (topical): all the rest (and probably the first also) are contrastive" (lbid, p. 48). The following is an example: (57) watakushi wa tabako wa suimasu. l TOP cigarette CON smoke 'Speaking of myself, I do smoke cigarettes.‘ 49 In this example, the first we is a topic marker, and the second is a contrastive marker. Kuno further notes that this sentence sounds incomplete in isolation; the listener expects to be given some statement that contrasts with smoking (Ibid, p. 49). Finally, ga rather than wa must be used with WH-words; (58) *Dare wa kimashita ka Who TOP came QM 'Who came?‘ (59) Dare ga kimashita ka Who SUBJ came QM 'Who came?’ (Kuno, 1973, p. 37) (Note: OM is a question marker) This situation can be easily explained if it is assumed that, like Chinese, Japanese WH-words, as well as TOPIC, move into COMP. If two things, TOPIC and a WH-word, were both in COMP, this would create a branching node which would interfere with c-commanding. To summarize, this section has shown that the analysis Huang proposes for Chinese is also relevant to Japanese. In particular, it has been shown that the TOPIC marker wa obeys constraints with respect to the SSC and the CNPC because Subjacency is violated, and the WH-lsland Constraint and a situation with two topic markers, because it creates a doubly-filled COMP condition. 50 Summary In keeping with other current studies which have postulated movement in Japanese at LF, this section has attempted to show how the TOPIC marker wa might be analyzed as moving to COMP position in LF. The results of this analysis show that wa obeys Subjacency in complex NP's and sentential subjects. Furthermore, this analysis also accounts for the fact that two topic markers cannot appear in the same sentence, (except in VP complements) and that we cannot appear in an NP with a WH-word. Thus, it would appear that this analysis is more than just an attempt to force the language to fit the mold. Since it helps account for several constraints in Japanese, it can be viewed as positive evidence that topic noun phrases occupy COMP position. IV. TOPIC CONSTRUCTIONS AND CASE Introduction Although a large number of postpositional particles exist in Japanese, only three are traditionally analyzed as case markers. They are ya (nominative), o (accusative) and ni (dative). (Ostler (1980), Kuno (1976), and Farmer (1984)). Various analyses of how case is assigned to these markers will be discussed in this section, as well as possible implications for the topic marker wa. A. Kuno's Analysis According to Kuno (1976), some particles are "in deep structure and some are inserted by transformations” (p. 328). He places ya, a and ni in the latter category. In other words, (60) would have the deep structure shown in (61): (60) John ga Mary ni okane o yatta. SUBJ IO money OBJ gave 'John gave money to Mary.‘ (Note: lO-lndirect Object) 51 52 (61) [John] Np [Marlep [okane]Np [yattaJV In support of this analysis, he cites the fact that the markers change when the sentence undergoes passivization. Sentences (62) and (63) are corresponding active and passive sentences: (62) John ga Mary 0 korosita SUBJ OBJ killed 'John killed Mary.’ (63) Mary ga John ni korosareta. SUBJ by was killed 'Mary was killed by John.‘ This analysis is in accordance with government-binding theory which states that corresponding active and passive sentences have the same deep structure (Van Reimsjick and Williams, 1987). The problem with Kuno's case-marking transformation rules is that they lack generality. According to Kuno (1976), the indirect object marker ni should be attached to "the second of three unmarked NP's that do not yet have a particle,” ge attaches to 'the first subject NP," and o to 'the first nonsubject unmarked NP to the left of the main verb if it is [-stative], and ya if it is [+stative]." (p. 330). These rules contain a lot of specific detail which, in the interests of achieving generality, it would be desirable to omit. However, it may be that Kuno's aim is descriptive, not explanatory. In addition, Kuno 53 has nothing to say about the topic marker wa. N B. Farmer's Analysis Farmer (1984) provides an account of case-marking which is more general and also deals with the topic marker wa. In her analysis, the PAS (predicate argument structure) is the main point of reference. Each predicate will have argument requirements associated with it. Thus, ”an intransitive predicate has only one argument slot, a transitive predicate has two, and a ditransitive predicate has three.” Like Kuno, the NP's which can be assigned case by the verb are ya, a and hi (pgs. 47-48). Application of case takes place through evaluation: Evaluation is a process that mediates between the syntax and the PAS. Its purpose is to associate an argument position with an overt N' that is a sister to the verb. The N'-ga in (2.69a) evaluates the GA argument position in (GA aruk). This evaluation is indicated by the use of indices i, j, k, etc. . . . these are not "referential" indices. (lbid, p. 49) 54 (64) V! NP-gai V aruk (GA; aruk) (lbid) In this example from Farmer, the verb aruk, 'walk,‘ assigns an argument position to the NP-ga through evaluation. Farmer explains that, ”the case-marked noun phrase can be in any position to the left of the verb . . and the argument slots do not have to be in any particular order.” (lbid). Apparently this analysis takes into account any instances of scrambling. The next example deals with the evaluation of an NP-wa phrase. According to Farmer, in a sentence without an NP-ga phrase, an NP-wa phrase may be associated with the nominative slot through indirect evaluation. (Direct evaluation may be defined as associating an NP with an argument from the PAS; indirect evaluation may be defined as coindexing an NP with the verb.) In other words, in (65), we may assume that Naomi-wa is given the NP-ga argument. 55 (65) /VI\ N' N' V Naomi-we udon-oi tabe-ta (GA Qi tabe) (Ibid, p. 85) (Naomi ate noodles) In (66) there is both an NP-wa and an NP-ga. After associating the NP-ga with the one argument in the PAS, there is no argument left for the NP-wa phrase. Farmer's solution is to coindex this NP-wa with the verb through indirect evaluation: (66) V' N' N' Vj Taroo-waj Hanako-gai iede-si-ta (EA; ide-si) (lbid, p.86) As for Taro, Hanako ran away from home. 56 This poses two questions. First is the problem that the NP-wa still has no case, but has simply been tied into the argument structure. It may be questioned whether this is desirable. If a topic marker is analyzed in COMP position as in section III, this implies that it has a domain over the entire sentence, both subject and predicate. Coindexing it with the verb does not capture that dominance. This is in line with Kitagawa (1988) who states that ”We indicates that the immediately preceding X' is outside of the domain of 'evaluation' in terms of the PAS of the nucleus” (p. 184). Second is the question of exactly what 'indirect evaluation” is. The best Farmer can do at defining it is to call it a form of "co-indexing.” If a more concrete definition cannot be given, perhaps this is just an ad-hoc device. Farmer's analysis is more inclusive than Kuno's in that it incorporates the topic NP and attempts to tie it into the predicate. However, what exactly this position is, and how it is arrived at, are only vaguely described. c. Kitagawa's Analysis In Kitagawa (1988), an attempt is made at'a more explanatory solution to the problem of case-marking the topic. The phenomenon 57 which he refers to as ”Topic Binding“ may be defined as follows: "The Topic X' must be bound pragmatically to an X' which is in the domain of the predication.” Furthermore, tapic in Japanese is defined as 'X'-wa, where X is [-V]." (p. 184). Thus, Kitagawa analyzes (65) as follows: (67) [Top [Ni Naomij-wa [predrl'op [N'i udon] wa] [Pred [N'i Pro [ Ni Pro [vtabeta] ] ]. 'Naomi ate noodles.’ (lbid) Unlike Farmer, Kitagawa does not allow NP-wa to be associated with the unassigned NP-ga argument, but instead analyzes the subject slot as being filled by Pro. The topic is then bound to Pro. This is further support for this paper's analysis that NP-wa does not simply move to the subject slot in the absence of an NP-ga. Summary In conclusion, Kitagawa's analysis is most preferred because it offers an analysis which can lead to universal distinctions by further clarifying the topic/subject contrast. First, it makes a 58 distinction between topics and subjects even in sentences which lack an NP-ga phrase by not requiring that the NP-wa phrase be moved into subject position. Second, it allows topic constructions to be evaluated at the X' level, which allows them to remain in a position of dominance over the subject and predicate. CONCLUSION The topic construction in Japanese has created a large amount of sometimes contradictory information. This may be due in part to the fact that traditional linguistics is biased toward western languages, which do not exhibit topic-prominent sentences. In order to provide a adequate account of topic constructions in Japanese, it is necessary to recognize the fact that the typology of Japanese is both subject-prominent and topic-prominent. This typology is supported by a large number of distinctions (Section II). Such an analysis also makes possible the fact that the topic marker we may be analyzed as moving to a position of dominance in a clause. Case-assigning is also clarified through this typology, because it allows a distinction to be made between subjects and topics. This viewpoint cannot solve all problems related to topic constructions in Japanese. First of all, it may be argued that the idea of movement is a useful analogy but a poor analysis, since, Japanese does not show overt movement the way a language like English does. It could be that a lexical-functional analysis, which is not configurationally-oriented like government-binding theory, may ultimately provide a more appropriate explanation. As regards 59 60 case-marking, the solution of ”pragmatic” evaluation is still a weak one; the explanatory question remains unsolved. However, this analysis is valid in that it applies the topic construction to aspects such as Subjacency and subject/topic typology, which are general parameters. In the sense that topic constructions can be put on a more generalizable level, the analysis is a useful one. BIBLIOGRAPHY Chafe, Wallace L. (1976). Givenness, Contrastiveness, Definiteness, Subjects, Topics and Point of View. In Charles N. Li (ed.), Subject and Topic. New York: Academic Press Inc. Chomsky, Noam. (1965). Aspects of the Theory of Syntax. Cambridge, MA: The,M.I.T. Press. Chomsky, Noam. (1980). Rules and Representations. New York: Columbia University Press. Crystal, David. (1985). A Dictionary of Linguistics and Phonetics. New York: Basil Blackwell Ltd. Darden, Lindley. (1991). Theory Change in Science: Strategies From Mendelian Genetics. New York: Oxford University Press. Dubois, Jean et. al. (1973). Dictionnaire de Linguistique. Paris: Librairie Larousse. Farmer, Ann Kathleen. (1984). Modularity In Syntax: A Study of Japanese and English. Cambridge, MA: The MIT. Press. Farmer, Ann Kathleen et al. (1986). A Note on Weak Crossover in Japanese. Natural Language and Linguistic Theory. 6: 503-526. Fukui, Naoki. (1988). LF Extraction of blaze: Some Theoretical Implications. Natural Language and Linguistic Theory. 4: 33-42. Hale, Ken. (1983). Warlpiri and the Grammar of Non-Configurational Languages. Natural Language and Linguistic Theory. 1:5-47. Huang, C.T. James. (1981). Move WH in a Language Without WH Movement. The Linguistic Review. 1: 369-416. Josephs, Lewis S. (1976). Complementation. In Masayoshi Shibatani (ed.), Syntax and Semantics Volume 5: Japanese Generative Grammar. New York: Academic Press Inc. 61 62 Keenan, Edward L. (1976). Towards a Universal Definition of ”Subject." In Charles N. Li (ed), Subject and Topic. New York: Academic Press Inc. Kitagawa, Chisato. (1982). Topic Constructions in Japanese. Lingua. 57: 175-214. Kuno, Susu mu. (1973). The Structure of the Japanese Language. Cambridge, MA: The MIT. Press. Kuno, Susumu. (1976). Subject Raising.‘ In Masayoshi Shibatani (ed.), Syntax and Semantics Volume 5: Japanese Generative Grammar. New York: Academic Press Inc. Kuroda, S. -Y. (1976). Subject. In Masayoshi Shibatani (ed.), Syntax and Semantics Volume 5: Japanese Generaa've Grammar. New York: Academic Press Inc. Kuroda, S.-Y. (1979). Generative Grammatical Studies in The Japanese Language. New York: Garland Publishing Inc. Kuroda, S.-Y. (1988). Whether We Agree or Not: A Comparative Syntax of Japanese and English. In William J. Posner (ed.), Papers from the Second International Conference on Japanese Syntax. Stanford, CA: Center for the Study of Language and Information. Lehmann, WP. (1976). From Topic to Subject in Indo-European. In Charles N. Li (ed), Subject and Topic. New York: Academic Press Inc. Li, Charles N. and thompson, Sandra A. (1976). Subject and Topic: A New Typology of Language. In Charles N. Li (ed.), Subject and Topic. New York: Academic Press Inc. McCawley, James (1976). Relativization. In Masayoshi Shibatani (ed.), Syntax and Semantics Volume 5: Japanese Generative Grammar. New York: Academic Press Inc. Nakau, Minoru. (1973). Sentential Complementation in Japanese. Tokyo: Kaitakyusha Co., LTD. 63 Ostler, Nicholas. (1980). A Theory of Case Linking and Agreement. Bloomington, IA: Reproduced by the Indiana University Linguistics Club. Pei, Mario. (1966). Glossary of Linguistic Terminology. Garden City, NY: Doubleday and Company Inc. Rutherford, William E. (1983). Language Typology and Language Transfer. In Susan M. Gass and Larry Selinker (eds), Language Transfer and Language Learning. Rowley, MA: Newbury House Publishers Inc. Saito, Mamoru and Hoji, Hajime (1983). Weak Crossover and Move Alpha in Japanese. Natural Language and Linguistic Theory. 1: 245-259. Sells, Peter. (1985). Lectures on Contemporary Syntactic Theories. Stanford, CA: Center for the Study of Language and lnforrnation. Van Riemsdijk, Hank and Williams, Edwin. (1986). Introduction to the Theory of Grammar. Cambridge MA: The MIT. Press. Yamag iwa, Joseph K. (1942). Modern Conversational Japanese. New York: McGraw Hill Book Company, Inc. "IIIIIIIIIIIIIIIIII“