ABSTRACT A GENERATIVE-TRANSFORMATIONAL STUDY OF NEGATION: A CONTRASTIVE ANALYSIS OF JAPANESE AND ENGLISH By Kunihiro Iwakura The present thesis is an attempt to formulate an analysis of negation that is applicable to both Japanese and English within the framework of the theory of generative- transformational grammar. To date, this remains to be done in that neither the interpretive nor the generative- semantic analysis of negation is satisfactory, viewed from the standpoint of its applicability to both Japanese and English. This work starts with a brief review of the recent 'hensformational works on negation in English such as Jackendoff (1969), (1971), Partee (1970), G. Lakoff (1969). U970a), (1970b), (1971a), and Garden (l970a), (1970b), Imting that they leave a number of problems unsolved. Thus, 'Umir analyses are far from satisfactory, apart from their inapplicability to Japanese. In Chapter II, the interrelations of negation and qmumifiers are discussed in detail, based on the two puta- tive universals regarding the sc0pe of negation and that of quantifiers: (l) the scope of negation is the whole sentence in which it occurs, (2) the sc0pe of a quantifier is the whole sentence in which it occurs. In particular, Kunihiro Iwakura zan analysis of negation and quantifiers is prOposed first for Japanese and then it is applied to English to test its validity. The proposed analysis involves several transfor— mations such as Sentence-raising, Negative-attachment, Quantifier-attachment, Adverbial—movement in both Japanese and English, and Contrastive Ea-attachment in Japanese, which are all shown to be independently motivated transformations. Moreover, a twofold distinction is made for negation, sentential and verb-phrase, and also for quantifiers, sentential and nominal. Justification for the twofold distinctions is given in a number of ways. In addition, the prOposed analysis is tested against a number of remaining problems to demonstrate that it can solve those problems. In Chapter III, the interrelations of negation and four classes of adverbials are discussed in detail. Noting similar behaviors of these adverbials to quantifiers with respect to negation, we apply the analysis prOposed in Chapter II to these adverbials. This application confirms the validity of the analysis. Additionally, another candi- date for a language universal is proposed regarding manner adverbials with respect to negation: manner adverbials cannot co-occur with verb-phrase negation. In Chapter IV, the topic of Negative-raising is discussed and it is demonstrated that a minor rule such as Negative-raising may be diSpensed with, or, to be more Kunihiro Iwakura exact, it may be incorporated into the general rule of Negative-attachment in our analysis. Furthermore, it is shown that the prOposed analysis can account for a number of cases that cannot be explained in terms of Negative— raising, namely, those cases where sentences of the type "NP think(s) [ X - not - Y 33" and their counterparts of the type "NP do(es) not think [ X — Y is" are different in meaning or grammaticality: this analysis can explain them in the same way as those cases where sentences of the two types are synonymous, in terms of the twofold distinction of negation. Thus, the proposed analysis is shown to have several advantages over the Negative-raising analysis. A GENERATIVE—TRANSFORMATIONAL STUDY OF NEGATION: A CONTRASTIVE ANALYSIS OF JAPANESE AND ENGLISH By Kunihiro Iwakura A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Linguistics and Oriental and African Languages 1973 To Seiko ii ACKNOWLEDGMENTS I am grateful to all the members of my Doctoral Committee for their very useful comments and suggestions. My deepest gratitude goes to Professor Seok C. Song whose ipatient guidance and warm encouragement made this disserta— tion possible. I am also greatly indebted to Professor Julia S. Falk for her extremely valuable comments. My sincere thanks also go to Professor Ruth M. Brend for her very helpful suggestions. Also, I am indebted to Professor John B. Eulenberg and Professor John T. Ritter for their helps during the preparation of this dissertation. I wish to take this Opportunity to express my grati- tudes to Dr. Donald J. Calista, Mr. & Mrs. John B. Brattin and Mr. John R. Barnett, who were kind enough to read the earlier drafts of this thesis and provided me with useful suggestions concerning the usage of English. My gratitude also goes to the Department of Linguistics and Oriental and African Languages at Michigan State Univer- sity for giving me a two-year teaching assistantship and to the Fulbright Commission for the travel grant, which enabled me to initiate my doctoral work in the United States. Finally, I should like to thank my wife, who performed the task of typing this thesis as well as its earlier drafts. iii CHAPTER I. FOOTNOT CHAPTER I 2.1. 2.2. 2030 204. 2.5. 2.6. 2.7. 2.8. 2090 2.10. 2.11. TABLE OF CONTENTS INTRODUCTION . . . . . . . . . . . . . . ES 0 O O O O O O O O O C O O O O O O I. INTERRELATIONS OF NEGATION AND QUANTI— FIERS I C O O O O O O O O O O O O O O O 0 Differences between Japanese and English Negation . . . . . . . . . . . . . . . . Negation and Sentential Hierarchy in Japanese Negation and Contrastive ya in Japanese . . . Quantifiers and Sentential Hierarchy in Japanese . . . . . . . . . . . . . . . . . . Quantifiers and Contrastive ya in Japanese . A PrOposed Analysis of Negation in Japanese . The Contrastive Ea Versus the TOpic HQ Attached to Quantifiers . . . . . . . . . . . The Application of the Proposed Analysis to English Negation and Quantifiers . . . . . A Proposed Analysis of Quantifiers with Respect to Negation . . . . . . . . . . . . 2.9.1. A PrOposed Analysis of Quantifiers in Japanese . . . . . . . . . . . . . . 2.9.2. The Application of the Proposed Analysis to English Quantifiers . The Discussion and Solution of Remaining PrOblems O O O O O O O O O O O O O O O 0 Conclusions . . . . . . . . . . . . . . . . FOOTNOTES . . . . . . . . . . . . . . . . . . . iv 12 13 14 18 BO 35 4o 42 68 74 99 99 122 131 157 159 V CHAPTEEi III. INTERRELATIONS OF NEGATION AND ADVER— 3.1. 3.2. 3.3. 3.4. 3.6. BIALS . . . . . . . . . . . . . . Negation and Nominal Adverbials in Japanese . Negation and Adverbials of Frequency in Japanese . . . . . . . . . . . . . . . . Negation and Adverbials of Reason and Adver- bials of Purpose in Japanese . . . . . The Application of the Analysis to the Corre- sponding English Adverbials . . . . . . . 3.4.1. Negation and Nominal Adverbials in English . . . . . . . . . . . . . . . 3.4.2. Negation and Adverbials of Frequency in EIlgliSh O O O O O C O O O O O O O 0 3.4.3. Negation and Adverbials of Reason and Adverbials of Purpose in English . Negation and Manner Adverbials . . . . . 3.5.1. Negation and Manner Adverbials in Japanese . . . . . . . . . . 3.5.2. Negation and Manner Adverbials in EngliSh O O O O O O O O O O O O O O 0 conCluSions O O O 0 O I O O O O 0 O O O O FOOTNOTES . . . . . . . . . . . . . . . . . . . . CHAPTER IV. NEGATIVE RAISING . . . . . . . . . . . . 4.1. 4.2. 4.3. 4.4. 4.5. Negative—Raising Versus the Proposed Analysis in Japanese . . . . . . . . . . . . . . . Advantages of the Proposed Analysis in Japanese . . . . . . . . . . . . . . . . . . The Application of the Proposed Analysis to English with Respect to Negative-Raising . . . Advantages of the Proposed Analysis in English Conclusions . . . . . . . . . . . . . . . . . FOOTIJOTES C O I O O O O O O O O O O O O O O O O O O 175 175 188 195 207 208 218 226 236 236 246 251 254 274 275 292 328 337 365 367 vi CHAPTER V. CONCLUSIONS . . . . . . . . . . . . . . 376 BIBLIOGRAPHY..................... 386 S-T Th VP ABBREVIATIONS Adjective Sentential Adjective Adverbial Degree Adverbial Adverbial of Frequency Location Adverbial Manner Adverbial Nominal Adverbial Adverbial of Reason and Purpose Time Adverbial Auxiliary Determiner Noun Negative Noun Phrase Quantifier Sentence Verb Verb of Saying and Thinking Verb of Thinking Verb Phrase vii CHAPTER I INTRODUCTION There have been a number of transformational works done in the topic of negation and quantifiers in English, notably, Jackendoff (1969), (1971), Partee (1970), G. Lakoff (1969), (1970a), (1970b), (1971a) and Garden (1970a), (1970b). These recent works have established that interre- lations of negation and quantifiers are related significantly to the semantic interpretation of a sentence in which they occur. These discussions result in part from the controversies between the interpretive (e.g. Jackendoff, Partee) and the generative semanticians (e.g. G. Lakoff, Garden). In this thesis we will discuss the same topic from a different viewpoint and with the aim of formulating an analysis of negation that is applicable to both Japanese and English. To date, this remains to be done in that neither the interpretive nor the generative-semantic analysis of negation is satisfactory, viewed from the standpoint of its applicability to both Japanese and English. To clarify the point, let us review briefly the recent transformational works on negation and quantifiers. 1.1. THde Interpretive Versus the G enerat ive-S emant ic Analysis Jackendoff and Partee try to account for the semantic interrelation of negation and quantifiers in terms of semantic interpretation rules. Jackendoff, in particular, resorts to the surface order of negatives and quantifiers to give the semantic interpretation of a sentence in which they occur. To cite an example from Jackendoff (1969) with a slight change, consider the following: (1.1) Nan many arrows hit the target. (1.2) HEQX arrows did nan hit the target. He argues that sentence (1.1), but not (1.2), is synonymous with "It is nan so that many arrows hit the target," and that this semantic difference is due to the difference in the surface order of nan and many: in (1.1) nay precedes many, while in (1.2) many precedes nan. On the other hand, G. Lakoff and Garden argue that negation and quantifiers are generated in the base as verbs of higher sentences, and that the semantic difference between (1.1) and (1.2) may be explained in terms of the hierarchical difference between 223 and many in the underlying structure: in the structure underlying (1.1), 222 occurs in a higher sentence than many, while the opposite is the case with that underlying (1.2). According to them, therefore, sentences (1.1) and (1.2) are derived from distinct underlying structures such as (1.3) and (1.4), respectively: (1.3) NP////,S r‘\\\\\\\‘w it/ N1D>S VIP M-\\\\\\\ Mt ‘P many arrows arrows hit the target (1.4) NPS/ lv\ NP////’ m‘\\\\\‘s mj:y l NPS/ 2\VP arrows it /sj\mt arrows hit the target The derivation of sentence (1.1) from (1 3) or (1 2) fITnn(1.4) requires the application of what G. Lakoff calls kQuaintifier-lowering. This transformation inserts quanti- jflers (probably negatives as well)1 into a lower sentence, 3for instance, many and not of (1.3) into 83’ one at a time. 1u2. Problems 9: the .Jnerpretive and the .Jmmrative-Semantic Analyses Lon31dering their analyses, we note that there are Emveral problems yet to be solved in both of the analyses. The analys1s in terms of the surface order of negatives and 4 mmntifiers must face the following problems: A. If one adopts the interpretive analysis based on the surface order of negatives and quantifiers, then moving transformations such as Passivization and Topicalization cannot remain meaning-preserving. For instance, observe the following sentences cited from Jackendoff (1969) with a slight change: (1.5)(=l.1) Nan many arrows hit the target. (l.6)(=1.2) Many arrows did nan hit the target. (1.7) The target was nan hit by many arrows. ‘We note that there is no reason to block the application of Passivization to the structure underlying sentence (1.6) to derive sentence (1.7). But the trouble is that sentence (1.7) is not synonymous with (1.6), so Passivi- zation in this case is not meaning-preserving. In turn, consider the following sentences cited again from Jackendoff (1969): (1.8) The police did nay arrest many demonstrators. (1.9) Many demonstrators were nay arrested by the police. (1.10) Na: many demonstrators were arrested by the police. If Passivization applies to the structure underlying (1.8), the resulting sentence is (1.9), not (1.10), in Spite of the fact that (1.8) is synonymous with (1.10), not (1.9). Thus, Passivization in this case is not meaning-preserving, either. Furthermore, one cannot v... e... r 5 derive the active counterpart of (1.10): if (1.8) and (1.9) are related by the Passivization transformation, what active sentence is related to (1.10)? One might argue, following Klima (1964), that (1.10) is derived from (1.9) by the application of what Klima (1964) calls Negative-incorporation which incorporates nan in (1.9) into many. But the trouble is that Negative-incorporation is not a meaning-preserving transformation since sentences (1.9) and (1.10) are not synonymous. To give up the meaning-preserving condition on 'transformations is clearly to increase the descriptive ,power of transformations, which only contributes to making Inore serious the defect of the theory of transformational grammar . 2 B. There are several types of counter—examples, as G. Lakoff (1969) notes, to the interpretive analysis in terms of the surface order of negatives and quantifiers. The first major type follows: (1.11) The arrows that did nan hit the target were man . In sentence (1.11), G. Lakoff argues, nan precedes many but (1.11) is synonymous with (1.6), not with (1.5). G. Lakoff notes that in (1.11) many follows nan but is in a higher sentence than nan, and this "asymmetric command relationship" between nan and many marks the difference in sc0pe. The second major type involves extra heavy stress, as in (1.12): (1.12) The students did 22: read many books. In sentence (1-12).£2£ precedes many, but the semantic interpretation of (1.12) is that in which many includes nay in its scope. Therefore, the analysis in terms of the surface order of negation and quantifiers must be revised to account for these counter-examples. On the other hand, the analysis in terms of the Quantifier-lowering rule proposed by G. Lakoff (1969) still leaves the following problems unsolved. (3- Quantifier-lowering, as criticized by Chomsky (1972:184-185), violates the presumably universal constraint that no rule may introduce an item into a jphrase of an embedded sentence from outside of that sentence. 13- The underlying structures in which negatives and Quantifiers occur as verbs of higher sentences, such as (1.3) and (1.4), are not well-motivated but rather an nan, as noted by Chomsky (1972). For instance, the sentence "Many men read few books" will have the following underlying structure with unnecessary details aside: (1.13) NP/ \my .a/ \ 16/ Sew\f 0/? \s / S\ men read books Chomsky (1972, 183), in particular, notes: Notice first that the structures in which quanti— fiers appear as predicates have unique properties. For example, the structure (79) [i. e. (1.13) with the most deeply embedded S missing] is admissible only if the embedded NP books has a relative clause attached to it; furthermore, this relative clause must contain both of the NP' s that appear in (79). These condi- tions are without parallel among syntactically motivated structures. Furthermore,a although (79) appears to involve a "relative clause", this structure is unique in that its antecedent, man, does not appear within the "relative clause" of which it is the antecedent. ]~- 3. Aims a: the Present T11 —.lesis From the brief review of the recent transformational ‘WOrks on negation and quantifiers, we see that both the interpretive and the generative-semantic analyses are far fibm satisfactory in that they leave several problems Unsolved. Moreover, from the viewpoint of its applicability to :bpanese as well as English, we may point out one more (fisadvantage of the interpretive analysis in terms of the surface order of negatives and quantifiers. In Japanese, 8 nmmtion occurs in the final position of a sentence, as will bediscussed in the subsequent chapters, with the result Mum negation must always follow quantifiers in their inuface order. Now, the point is clear: the interpretive analysis in terms of the surface order of negation and quantifiers is completely inapplicable to Japanese negation and quantifiers. Our aim in this connection is, therefore, 'to explore and propose an analysis of negation that will be appfljcable to both Japanese and English negation, in Ehidition to being free from the above criticisms. What is the significance, then, of the applicability (If our analysis to both Japanese and English? In the theory 0f7 transformational grammar, one of the fundamental goals is PC) construct a linguistic theory of explanatory adequacy WTrich makes it possible to select a descriptively adequate Elfiimmar for each language. The problem is, therefore, to (“instruct a general theory of language which is so richly EstI‘uctured that it can sufficiently narrow the class of ZPOSsible grammars so that the problem of selecting a grammar <3an be approached. Our research should be, therefore, along 'the lines of enriching the general theory and simplifying ‘Particular grammars.. Viewed from this standpoint, a rule aDplicable to two languages is more significant than a rule amflicable to a particular language. The reason is obvious: arule applicable to two languages is more likely to be a language universal which may contribute to enriching the gmneral linguistic theory and simplifying particular 9 mammars. Moreover, if a rule is applicable to two lmmmages which are genetically unrelated, it is more likely totm a language universal than a rule applicable to two genetically related languages. Here lies the significance of the applicability of our analysis to both Japanese and English. Japanese and English are genetically as unrelated as any two languages in the world. Thus, an analysis appflicable to both Japanese and English is very likely or at Ileast more likely to be applicable to other languages than all analysis applicable to English alone. Furthermore, some <3f the rules or constraints contained in such an analysis Eire very likely to be language universals. To repeat, the aim of the present dissertation is to Cliscuss several aspects of negation and present an analysis (If negation that is applicable to both Japanese and English. 1-4. The Outline at; the P\rgsent Thesis The outline of this thesis is as follows. In Chapter II, I discuss the interrelations of negation and quantifiers. The discussion is based on an observation 'Which seems to be a linguistic universal regarding the scope of negation: the scope of negation is the whole sentence in Vmich it occurs. Then, I argue that if a constituent is not hmluded in the scope of negation, it must be outside the amntence containing the negation in the underlying structure. Ehsed on the sc0pe of negation, I also make a twofold dis- tinction of negation, sentential and verb-phrase; only 10 samential negation can include quantifiers in its sc0pe. bet, noting the similar behaviors of quantifiers to Ingatives, I go on to argue that the scope of quantifiers is 'Uuewhole sentence in which they occur. Then, if a constituent is outside the scope of a quantifier, it must be outside the sentence containing the quantifier in the ‘underlying structure. Additionally, I make a twofold dis- ‘tinction of quantifiers, sentential and nominal, parallel to ‘that of negation. Thus, the proposed analysis incorporates tvno putative universals regarding the scope of negation and ‘bhat of quantifiers along with the twofold distinction of negation and that of quantifiers. It also involves several tI‘ansformations which are shown to be independently IIIO‘tivated in both Japanese and English. Next, I demonstrate Inlat the proposed analysis can account for a number of facts COncerning the interrelations of negation and quantifiers 111 both Japanese and English, in addition to solving the ahove-noted problems. In Chapter III, I discuss the interrelations of negation and four classes of adverbials. I note first that those adverbials behave quite similarly to quantifiers in that, depending upon whether they co—occur with sentential (n‘verb-phrase negation, they show semantic differences like Tmntifiers. Then, applying the analysis proposed in Qmpter II to these adverbials with respect to negation, I demonstrate that the proposed analysis is general enough to cover these adverbials. In addition, I propose a presumably 11 mnversal constraint regarding manner adverbials: manner adwntials cannot co-occur with verb-phrase negation. If thhsconstraint is correct, I argue, it gives additional mofiyation to the twofold distinction of negation. In Chapter IV, the tOpic of Negative-raising is discussed within the prOposed framework. I attempt to demonstrate that a minor rule such as Negative-raising can 1m dispensed with, or put differently, it can be incorpo- rated into the more general rule of Negative-attachment in mu'auialysis. In addition, it is shown that the prOposed anakysis can account for a number of cases that cannot be explailmfl.in.terms of the Negative-raising rule, namely, those cases where sentences of the two types--the type "NP think(s) [ X - not - Y 38" and the type "NP do(es) not think [ X - Y JS"--are different in meaning or grammaticality. Furthermore, it can explain them in the same way as those Cases where sentences of the two types are synonymous, in terms of the twofold distinction of negation. Thus, the plmmmsed analysis is shown to have several advantages over the Negative-raising analysis. Chapter V is a brief recapitulation of the results of the thesis. 1. CHAPTER I FOOTNOTES G. Lakoff (1969), (1970b), (1971a) does not seem to claim explicitly that Quantifier-lowering inserts .negatives as well into lower embedded sentences. If the zrpplication of Quantifier-lowering is restricted to (Iuantifiers, some other transformation is necessary to ilasert negatives into lower embedded sentences. In this connection, see Chomsky (1972:124-125) who, in Particular, notes: The gravest defect of the theory of transformational grammar is its enormous latitude and descriptive power. . . . Virtually any imaginable rule can be described in transformational terms. Therefore a critical problem in making transformational grammar a substantive theory with explanatory force is to restrict the category of admissible phrase markers, admissible transformations, and admissible derivations. 12 CHAPTER II INTERRELATIONS OF NEGATION AND QUANTIFIERS Recent researchers, notably, Jackendoff (1969), (1971) and G. Lakoff (1969), (1970b), (1971a) have established that interrelations of negation and quantifiers are related Significantly to the semantic interpretation of a sentence hlvdlich they occur. In this chapter we will discuss the same tOpic from a different viewpoint and with the aim of formulating an analysis of negation that is applicable to bOVh Japanese and English negation with respect to quantifiers. We start by discussing Japanese negation and propose an analysis that may take care of a number of facts about JaPanese negation. Then, we will go on to apply this analy- sis to English negation to test and to show its validity. OIll‘arguments in this chapter involve, among others, the dimnmsion of the following hypotheses about Japanese and English: 1. The scope of negation is the whole sentence in which it occurs. 2. The constituents that are outside the sc0pe of negation occur outside the sentence containing the negation in the underlying structure. 13 14 3. There are two types of negation, sentential and verb- phrase: sentential negation, but not verb-phrase negation, "commands" the sentence it negates in the underlying structure. 4. There is a striking parallel observed between negation and quantifiers, and the latter may be treated in a way similar to negation. 5. {The scope of a quantifier is the whole sentence in which :it occurs. 6. {Tkmaconstituents that are outside the s00pe of a quanti- Ifier occur outside the sentence containing the (luantifier in the underlying structure. 7- Tlmme are two types of quantifiers, sentential and inominal: they will be derived by the rule that rewrites VP and the rule that rewrites NP, respectively. Both Japanese and English contain transformations such as Sentence—raising, Negative—attachment and Quantifier- attachment. In addition, Japanese grammar has a Contrastive ya—attachment transformation. 2.1. Differences between \Japanese and English \Negation Before we start our major discussion, it will not be lufit to note several striking differences between Japanese an English negation. Our interest is not in these dif- ibrences themselves but to show that in spite of these differences the analysis which we will prOpose later is l5 mmlicable to both Japanese and English negation. Some rfltvant differences between Japanese and English negation amaas follows: A. Japanese has no NP negation, nor what Klima (1964) calls Negative-incorporation by which the negative is incor- Ixnated into quantifiers such as anything + nag-——9 nothing. In Japanese, therefore, we cannot isolate NP negation such as "no student in the class" in the sentence ”No student in the class can answer the question." Its Jagnanese correspondent may be given only as a discontinu- ous string: (2L1) sono kurasu no dono gakusei mo . . . nail the class in any student not The correspondent of the whole sentence "No student in the class can answer the question" will be as follows: (2.2) sono kurasu no dono gakusei mo sono situmon hi the class in any student the question kotaerare nai. can-answer not Furthermore, we note that a more exact English corre- spondent of (2.2) is (2.3) rather than "No student in the Class can answer the question" in that "any . . . 221" corresponds to "gang . . . nan": (2.3) *Any student in the class cannay answer the question. Similarly, Japanese has no exact correspondent of (2.4) but only of (2.5): (2.4) I know nothing about it. 16 (2.5) I do not know anything about it. {Mm Japanese equivalent of (2.5) will be given as: (2.6) watasi wa sore ni tuite nanimo sira nai. I it about anything know not B. Japanese has no incomplete negatives corresponding to the English few, little, hardly, scarcely, seldom and rarely. Klima (1964) assumes that they contain a negative, noting that they show a number of syntactic similarities to mmnplete negatives such as not, never and nothin , esrmecially with respect to tag-question formation, SUIuject-auxiliary inversion and others. The validity of (His assumption is partly confirmed in the Japanese Counterparts of may, little, hardly, scarcely, seldom and Ififlflgly, which consist of "adverbial + negative." For instance, the Japanese counterpart of fan is "hotondono ° . . 2&1": notice that there is no English counterpart of hotondgno in this sense, since "hotondono . . . 2E1" as [a unit corresponds to the English fan. The same is true of the other adverbials in question. To clarify the point, let us consider the following examples, in which "hfligndgng . . . nag," ”hotondo . . . nai" and " mettani .. . nai" as units correspond to few, little and seldom, respectively, with the result that there is no English correspondent of hotondono, hotondo or mettani: (2.7) (2.8) (2.9) 17 nihon de wa hotondono gakusei ga ratengo 0 Japan in student Latin benkyoosi nai. study not 'Few students study Latin in Japan.‘ sono mondai ni wa hotondo tyuui ga haraware the matter to attention paid nakatta. not-was 'Little attention was paid to the matter.‘ kare wa mettani eiga ni ika nai. he movies to go not 'He seldom goes to the movies.‘ Moreover, Klima (1964) has observed that only has a number ofproperties in common with a negative and this observa— tionis confirmed by one of its Japanese correspondents COmposed of "adverbial + negative." For example: (2.10) sono syoonen sika kareno hahaoya 0 that boy his mother settoku—deki nai.2 persuade can not 'Only that boy can persuade his mother.‘ hlsentence (2.10), since "sika . . . nai" as a unit corre- Sponds to only, there is no English equivalent of sika. 0. If Japanese has verb-phrase negation like that in English, that is, negation which is associated with the verb or the verb phrase of a sentence, there is a striking difference between them: the negative precedes the verb in English, while in Japanese it follows the verb which 18 occurs sentence-finally in a simplex sentence. To illustrate with a concrete example, compare the following: (2.11) a. sono huzin wa sinsoo o sira nai. the lady truth know not 'The lady does not know the truth.‘ b. The lady does 22E know the truth. In (2.llb) nan precedes the main verb KEQE: while in its Japanese equivalent the negative 2%; follows the verb 3 m which usually occurs sentence-finally in an affirma- tiveesimplex sentence. The difference in the position Whexmathe negative occurs in a sentence is an important factxn*to be considered in the contrastive analysis of JaIuanese and English negation with respect to quantifiers, as will be discussed in sections 2.8, 2.9 and 2.10. $320 Negation and Sentential ‘3£33£221.12 Japanese In Japanese, negation occurs sentence-finally immedi- ately'following a verb in a simplex sentence. This position Ofnegation in a sentence has much to do with the sc0pe of negfiuon in Japanese. To illustrate with a concrete example, QOnsider the following: (2.12) kare wa kinoo tosyokan de benkyoosi nakatta. he yesterday library in study not-did 'He did not study in the library yesterday.‘ Phat, we observe that sentence (2.12) may be synonymous nth (2.13) in Spite of the clear difference between them: in (2.12) the negative occurs in a simplex sentence, while 19 in (2.2L3) the negative occurs in a higher sentence: (2.];3) [[kare wa kinoo tosyokan de benkyoosita wake]S he yesterday library in studied that de wa naiJS4 is not 'It is not so that he studied in the library yesterday.’ Second, we observe that the negation in (2.13) may negate any Of the following: (2.14) a. kare wa kinoo tosyokan de benkyoosita. he yesterday library in studied 'He studied in the library yesterday.‘ b. kare he c. kinoo yesterday d. tosyokan de library in e. benkyoosita studied ACCOI‘dirlgly, any of the following sentences may occur f“lowing sentence (2.13): (20135) a. kare wa nanimo si nakatta. he anything do not-did 'He did nothing.’ b. Mary ga benkyoosita. studied 'Mary studied.‘ 20 ototui sita. the day before yesterday did '(He) did the day before yesterday.’ ie de sita. home at did '(He) did at home.’ nemutta no da. slept that is 'Lit. It is that (he) slept. (= (He) did sleep.)' This observation shows that: (2.16) Negation in a higher sentence as in (2.13) includes in its sc0pe the whole sentence in which it occurs, including its lower sentence. Here, "inc1uded in the sc0pe of negation" means that nega- tion Huay negate not only the whole sentence as a unit but also ally constituent in the sentence, if it is included in the Scope of negation. FUrthermore, we observe that any of sentences (2.15a) tmr9W8T1 (2.15e) may occur following sentence (2.12). This Observation and the synonymity of (2.12) with (2.13) demons t rat e that: (2.17) Verb-phrase negation as in (2.12) may include in its scope the whole sentence in which it occurs. T he above (2.16) and (2.17) may be conjoined into: (2.18) Negation, whether verb-phrase negation or negation in a higher sentence, may include in its scope the whole sentence in which it occurs. 21 IFrom the foregoing discussion we see that verb-phrase rmgatixxn, though it occurs sentence-finally, may include in its sc0pe the preceding whole sentence as in (2.12). But this is not always the case: there are some constituents which are not included in the scope of verb-phrase negation in.Japanese. To illustrate with an example, let us consider the following: (2.19) kore wa mattaku tadasiku nai. this altogether correct not 'Lit. This is altogether not correct.’ Shun; verbs and adjectives occur sentence-finally (followed by negatives in negative sentences), adverbials must precede them. Incidentally, it is generally observed that Japanese word cxrder with respect to a verb and various kinds of adver- bials :is the mirror image of the corresponding English word Order. To illustrate the point, compare the following: (2-2C3) a. kanozyo wa [ototui [tosyokan an she the day before yesterday library in [issyookenmei [benkyoositallll hard studied b, She [[[Estudied] hand] _i_r_1 the library] the d_ax before yesterday] correspondingly, in (2.19) the adverbial mattaku "altogether" precedes the negative n__a_i_. The relevant dif- ference between (2.19) and (2.12) is as follows: (2.12) may be synonymous with (2.13), whereas (2.19) is not synonymous vfith (2.M21), though the grammatical relationship between (2°19) and (2.21) is parallel to that between (2.12) and 22 (2J13) : (2.21) [[kore wa mattaku tadasii wake]S de wa nai]S this altogether correct that is not 'It is not so that this is altogether correct.’ We see that in sentence (2.21), the negative nag in the higher sentence includes in its sc0pe the relevant remainder of the sentence, including mattaku "altogether." The non- Synonynfity of (2.19) with (2.21) indicates that the negative in (2.19) does not include mattaku in its scope. Thus, the adverbial mattaku is one of the constituents which are not huflnuied in the scope of verb-phrase negation. Continuing this discussion, let us proceed to Consider: (2.222) sono siyoonin wa taihen syooziki de nai. the employee very honest is not 'The employee is not very honest.’ or 'The employee is very dishonest.’ In Serrtence (2.22) the negative nag may or may not include 'Um kaJle sentence in its sc0pe: that is, sentence (2.22) is ambiguous. If the rest of the sentence is included in the scoPe of negation, it will be synonymous with (2.23), in which-‘the negative in the higher sentence includes the remainder of the sentence in its sc0pe: (2'23 ) [[sono siyoonin wa taihen syooziki na wakeJS the employee very honest is that de wa nails is not 'It is not so that the employee is very honest.‘ 23 On‘thta other hand, if the negative does not include the adverlxial taihen ”very" in its scope, the reading of (2.22) wiLl'be synonymous with that of (2.24) in which the adver- bial taihen includes the negative prefix nn "dis-, non-, un-" in its sc0pe: (2.24) sono siyoonin wa taihen husyooziki da. the employee very dishonest is 'The employee is very dishonest.’ Let us cite another example which is similarly ambiguous, depending on whether the negative includes the Whale sentence in its SCOpe or not: “2.25) sono siyoonin wa kimi yori syooziki de nai. the employee you more-than honest is not 'The employee is not more honest than you.’ or 'The employee is more dishonest than you.’ When time negative in (2.25) includes the remainder of the Senteruze, (2.25) is synonymous with (2.26): 02°265) ([sono siyoonin wa kimi yori syooziki na the employee you more-than honest is wake]S de wa nai]S that is not 'It is not so that the employee is more honest than you.’ 0 - therwlse, (2,24) is synonymous with (2.27): (2'277) sono siyoonin wa kimi yori husyooziki da. the employee you more-than dishonest is 'The employee is more dishonest than you.‘ FI‘om the foregoing discussion, it should be clear that: 24 l. 'VExrb-phrase negation may not include in its scope <3ertain kinds of adverbials such as mattaku "altogether," 'taihen."very" and . . . yori "more than . . . J' 2. §3entential hierarchy is used in Japanese so that negation may include these adverbials in its scope: in (2.21), (2.23) and (2.26), the negation in a higher sentence can include in its sc0pe the lower sentence containing the adverbial in question. .However, there are some cases in which the use of sentential hierarchy is unnecessary to avoid the ambiguity 0f Senrtences involving those adverbials in question. To ilhnstrate the point, observe the following sentence: (2.2fi3) sono siyoonin wa taihen husyooziki de nai. the employee very dishonest is not 'The employee is not very dishonest.‘ AS COIItrasted with (2.22), sentence (2.28) is unambiguous: its reading is that in which the adverbial taihen "very" is inchided in the sc0pe of negation. Thus, sentence (2.28) is %monyqnous with.(2.29), in which the negation in the higher SentenCe contains the adverbial in its s00pe: (2'25?) [[sono siyoonin wa taihen husyooziki na wake]S the employee very dishonest is that de wa nai]S is not 'It is not so that the employee is very dishonest.’ T he same is true of the following: 25 (2.230) sono siyoonin wa kimi yori husyooziki de nai. the employee you more—than dishonest is not 'The employee is not more dishonest than you.‘ That is to say, (2.30) is synonymous with (2.31), in which the negation includes the whole sentence in its s00pe: (2.31) [[sono siyoonin wa kimi yori husyooziki na the employee you more-than dishonest is wake]S de wa nai]S that is not 'It is not so that the employee is more dishonest than you.’ Comparing (2.28) with (2.22), and (2.30) with (2.25), Velmate that the relevant difference is the occurrence of QESytujziki "dishonest" in (2.28) and (2.30) versus syooziki "honest" in (2.22) and (2.25). Thus, this must be the cause <3f the above-observed difference that both (2.28) and (2-3CX) are unambiguous, while (2.22) and (2.25) are ambig‘L‘lcaus. A closer comparison of husyooziki "dishonest" With .Sxooziki "honest" shows that husyooziki "dishonest" cmntaills the negative prefix nn "dis-, non-, un--" YGt, evmliif‘ they include no negative affix, certain adjectives show behaviors similar to those of husyooziki "dishonest" mm otflmegr negative adjectives. For instance, consider the fOIIOwing: (2°32?) kareno ie wa taihen tiisaku nai. his house very small not 'His house is not very small.‘ 26 (2.233) kanozyo wa taihen minikuku nai. she very ugly not 'She is not very ugly.’ Wecdbserve that both (2.32) and (2.33) are unambiguous and synonynmus with (2.34) and (2.35). TGSPeCtiVGlY‘ (2.34) [[kareno ie wa taihen tiisai wake]S de wa nai]S his house very small that is not 'It is not so that his house is very small.’ (2.35) [[kanozyo wa taihen minikui wake]S de wa nai]S she very ugly that is not 'It is not so that she is very ugly.‘ II1(2.34) and (2.35) the negation clearly includes the adverfllial taihen ”very" in its sc0pe. The synonymity of (2.32L) and (2.33) with (2.34) and (2.35), respectively, indicxates that the negation in (2.32) and (2.33) also incltmles the adverbial in its scope. Now, the relevant dif- ferenrze between (2.32) and (2.33) on the one hand and (2.22) on thfs other is the occurrence of tiisaku "small" and W "ugly" versus syooziki "honest." Owing to this difference, (2,32) and (2.33) are unambiguous, while (2.22) is ambiguous. ~35rom the above observations we see that in (2.28), (2°30) , (2,32) and (2.33) expected ambiguity is not materi- Eflized- (Dwing to some factor. This factor must be reduced to some Property common to husyooziki "dishonest," tiisai "smallfit and minikui "ugly" which is not found in syooziki 'WmneStp. and other positive adjectives. It is important to lufifi hffire that we can talk of the degrees of being, for 27 instauace, syooziki ma 2&1 "not honest," and the higher degreme of being syooziki ma na; "not honest" may be equal to'beirmghusyooziki "dishonest." This is confirmed by the fact that other degree adverbials such as sukosi "a little" and kanari "rather" as well as taihen "very" and . . .pyori "more than . . . " may co—occur with syooziki aa nai "not honest." For instance, observe the following: (2.36) sono siyoonin wa sukosi [syooziki ma nai] the employee a little honest is not tokoro ga aru. something exist 'The employee has something a little dishonest about him.‘ In Corrtrast, there is no such degree of being husyooziki an 931 "Ilot dishonest," tiisaku nai "not small" or minikuku nai "HOt tugly"; one cannot say someone is very not dishonest or Somewtrat not dishonest. Thus, they cannot co-occur with degree, adverbials, as is shown by the ungrammaticality of the following: (2°37) *sono siyoonin wa sukosi [husyooziki $1.9. nai] the employee a little dishonest is not tokoro ga aru. something exist 'Lit. The employee has something a little not-dishonest about him.‘ ( 2°3E3) *kareno ie wa sukosi [tiisaku nai] his house a little small not 'Lit. His house is a little not-small.‘ We 0b s €rr"’ncan.include these adverbials in its scope without involving sentential hierarchy. finfiplifying this discussion, let us consider next the Ollowing sentence: (2. 4CD) kanozyo wa koohukuni sina nakatta. she happily die not-did 'She did not die happily.‘ 7 includes 29 the adverbial koohukuni "happily" in its scope. This observation is confirmed by the fact that (2.40) is synony- mous with (2.41), in which the negation in the higher sentence includes the adverbial in its s00pe: (2.41) [[kanozyo wa koohukuni sinda wake]S de wa nai]S she happily died that is not 'It is not so that she died happily.’ 0n the basis of this observation, sentence (2.40) may be semantically analyzed into: (2.442) [[kanozyo wa koohukuni sina] nakatta] she happily die not-did New,.lem us consider what factor makes sentence (2.40) umnflxigmous. Adverbials such as koohukuni "happily" are chtxi manner adverbials and they are used to describe the WQVCnae performs an action in the broad sense of the word in that "(Jne dies” is "one performs an action of dying." D(Murally, if an action is not performed, one cannot (uncrilae the way the action is performed. This very nature Offlmruier-adverbials has much to do with the factor that mErlkes sentence (2.40) unambiguous. If the negative in “L4O) (ices not include the adverbial in its s00pe, the adverbial must occur outside the sentence containing the negative , since the sc0pe of the negative is the whole sente . ' nee; Jln,WhiCh it occurs. In this case, therefore, Sent ence (2.40) will be semantically analyzed into: (2. 43;) [kanozyo wa koohukuni [sina nakattal] I she happily die not—did n (2.43 > ‘the manner adverbial includes the negative in its ~ 30 scope and is meant to describe the action sina nakatta "did. not die." But we see that sina nakatta is not an action which may be described by a manner adverbial since aina nakatta indicates that no action of dying happened. More generally, manner adverbials cannot describe the contents expressed by the "verb + negative" or put differently, manner adverbials cannot co-occur with a negated verb. This very nature of manner adverbials is the factor that excludes the interpretation of (2.40) in which the manner adverbial is not included in the scope of negation. Thus, sentence (2.40) is another example in which some factor excludes one Of the two interpretations so that negation may include certain adverbials in its scope without involving sentential hierarchy. Summarizing the preceding discussion, we have demonstrated that negation in Japanese may not always include oer158.2111 kinds of adverbials in its scope unless (1) it involVes sentential hierarchy, or, (2) some factor excludes One of the two interpretations in which the adverbials in que3131011 are not included in the scope of negation. 2'3: IVEB 't' d (3 t t‘ & . ‘ga lon an on ras lve 141 iamanese There are two kinds of ya in Japanese, topic E and Cont - rastlve ya. To illustrate with an example, consider the fOllOWing. 31 (22.44) Bill ya benkyoosite iru. studying is 'As for Bill, he is studying.‘ or 'Bill (in contrast with someone else) is studying.‘ In (2344) HQ may be interpreted as a topic ma or as a contrastive ma. Thus, sentence (2.44) is ambiguous, depending on which ya is involved. If ya in (2.44) is a topic ya, the reading of (2.44) will be "As for Bill, he is shxtying" and this is clearly the primary reading of (2.44). On‘the’other hand, if ya in (2.44) is taken to be a contzmistive ma, (2.44) can mean "Bill (in contrast with somecule else) is studying." In this case, sentence (2.44) implites that someone else relevant in the context is not stmtyitmy What concerns our discussion in this section is the lertter ma, namely, the contrastive one in connection with negation. 5P0 start with, consider the following sentence: (2-4f5) kore wa mattaku tadasiku ma nai. this altogether correct not 'This is not altogether correct.’ We observe that in sentence (2.45) the negative nai includes theeuivtarbial mattaku "altogether" in its scope and so it is Emmnynnglls with (2.46) involving sentential hierarchy: (2'46) (=2.21) [[kore wa mattaku tadasii wake]S de wa this altogether correct that is 118118 not 'It is not so that this is altogether correct.‘ 32 In (12.46) the negative nag occurs in the higher sentence, thus; including the adverbial in its SCOpe. Comparing (2.45) with (2.19), we note that the only diffflerence is the presence of ma in (2.45) and its absence hi (2.19), and this minor difference causes a clear semarmic difference between them: sentence (2.45) is synonymous with (2.46), whereas (2.19) is not. Continuing the discussion, let us consider next the following: (2.417) sono siyoonin wa taihen syooziki de ya nai. the employee very honest is not 'The employee is not very honest.‘ We obnserve that sentence (2.47) has only one interpretation invdmich the negative includes the adverbial taihen "very" in its sc0pe. This observation is confirmed by the fact that (2.47) .is synonymous with sentence (2.48), in which the Imgatjgye in the higher sentence includes the adverbial in its SCOpe: (2°4E3)(=2.23) [[sono siyoonin wa taihen syooziki na the employee very honest wake]S de wa nai]S that is not 'It is not so that the employee is very honest.‘ Let us cite another related example: (2. 45’) sono siyoonin wa kimi yori syooziki de ya nai. the employee you more-than honest is not 'The employee is not more honest than you.’ 33 Smularly, sentence (2.49) is unambiguous and is synonymous with (2.50), in which the adverbial is included in the scope of negation: (2.50)(=2.26) [[sono siyoonin wa kimi yori syooziki the employee you more—than honest na wake]S de wa nails is that is not 'It is not so that the employee is more honest than you.‘ We further note that the contrastive ya may be attached8 to the adverbial . . . yori "more than . . . " itself as in (2.51): (2.51) sono siyoonin wa kimi yori ya syooziki de nai. the employee you more-than honest is not 'The employee is not more honest than you.‘ That :is to say, sentence (2.51) may be synonymous with both (2-49) and (2.50). To take another similar example, consider the following: (2.532) sono siyoonin wa taihen husyooziki de ya nai. the employee very dishonest is not 'The employee is not very dishonest.‘ AS exPected, sentence (2.52) is also unambiguous. In this mmnecrttion, recall that sentence (2.52), even without the Contras‘tive ya, is unambiguous, as discussed in the preceding section. ‘A* Esimilar argument holds for the following sentence 1' 121701171 r1 - S a manner adverbial: 34 (2.53) kanozyo wa koohukuni sini wa si nakatta. she happily die do not-did 'She did not die happily.‘ finmence (2.53) has only one reading synonymous with that of (2.54) involving sentential hierarchy: (2.54)(=2.4l) [[kanozyo wa koohukuni sinda wake]S de she happily died that is wa nai]S not 'It is not so that she died happily.‘ Furthermore, it should be mentioned that the contrastive ya can be attached to the manner adverbial itself as in: (2.55) kanozyo wa koohukuni ya sina nakatta. she happily die not-did 'She did not die happily.‘ Sentenlce (2.55) is synonymous, as expected, with both (2.53) and (2.54). The foregoing discussion demonstrates that the oontzmastive ya is used in Japanese so that negation may hkflluie in its scope certain kinds of adverbials which are otheI‘Wise not included in the scope of negation. Recall that seantential hierarchy, as discussed in the preceding Section, is another means for this purpose. It Will be Of interest if these two means, namely, the contrastive W_a_ and Smmerrt1161l hierarchy with respect to negation, may be related “Dea011 c>ther by a general principle. If they turn out to be WC I‘ela-‘tSed, the synonymity of (2.45) With (2.46). (2.47) -Hh (23‘- <4u8), (2.49) and (2.51) with (2.50), and (2.53) ‘ 35 mm.(2.55) with (2.54) will be accounted for in a general way. We will discuss this topic in section 2.5. 2.4. Quantifiers and Sentential Hierarchy 13 Japanese Japanese quantifiers show significant similarities to those adverbials previously cited in that they are not included in the s00pe of the negation which appears no higher than the quantifiers. To illustrate with a concrete example, consider first the following: (2.56) ookuno hito ga sono sensoo o nozoma nakatta. many people the war want not-did 'Many people did not want the war.’ We<5bserve that in (2.56) the negative nann does not include the cplantifier ookuno "many" in its SCOpe; on the contrary, the quantifier includes the negative in its scope. Aocorciingly, (2.56) is not synonymous with (2.57), in which the rngative in the higher sentence includes the quantifier in its scope: (2-EV7) [[ookuno hito ga sono sensoo o nozonda wake]S de many people the war wanted that is wa nails not 'It is not so that many people wanted the war.‘ Ga t9 ‘tcg the contrary, (2.56) is synonymous with (2.58), in Whic h ‘txk1€3 quantifier in the higher sentence contains the ati . V763 in its scope: ‘ 36 (2.58) [([sono sensoo o nozoma nakatta]S hito]NP ga the war want not-did people ooku itals many were 'There were many people who did not want the war.’ It is important to note that sentence (2.57) is an mmmple showing that negation occurs in a higher sentence so that it may include in its sc0pe a quantifier that is (wherwise not included in the scope of negation. Consider next the following: (2.59) sono gakuseitati wa ookuno hon o yoma nakatta. the students many book read not-did 'Many books, the students did not read.‘ or 'The students did not read many books.‘ werlo‘tethat sentence (2.59) is ambiguous with two readings, depondiing on whether the quantifier ookuno "many" is inCIUGted in the SCOpe of negation or not. More Specifically, it may be synonymous with either of the following: (206C)) (IIsono gakuseitati ga yoma nakatta]S hon]NP ga the students read not-did book ooku atta] many were S 'There were many books that the students did not read.‘ (2'61-) [[sono gakuseitati wa ookuno hon o yonda wake]S the students many book read that de wa nai]S is not 'It is not so that the students read many books. (= The students did not read many books.)' 37 It may be observed that in (2.61) the negative in the higher sentence includes the quantifier in its sc0pe, while the (mposite is the case with (2.60). (Thus, sentence (2.59) may occur in either of the following contexts: (2.622) sono gakuseitati wa ookuno hon o yoma nakatta, the students many book read not—did nazenara izuremo muzukasikatta node. the reason—was all difficult-were because 'Many books, the students did not read, the reason was that all of them were difficult.’ (2.635) sono gakuseitati wa ookuno hon o yoma nakatta the students many book read not-did ga sukunakutomo nansatukano hon o yonda. but at least some book read 'The students did not read many books but read at least some books.‘ Tflle point here is that sentence (2.61) is another exmmflfié in which negation occurs in a higher sentence so thatitmay include a quantifier in its scope. AllIPlifying this discussion, we may go on to examine the follOWing: (2'64) sono kurasu no subeteno gakusei ga sono sensei the class in all student that teacher 0 sonkeisi nakatta. respect not-did 'All the students in the class did not reSpect _ that teacher.‘ ”'8 0b SeIMVTe that in (2.64) the quantifier subeteno "all" is outSid e 'tlle sc0pe of negation. This observation is 38 confirmed by the fact that (2.64) is not synonymous with (2.65), in which the quantifier is inside the scope of the rwgative in a higher sentence: (2.65) [[sono kurasu no subeteno gakusei ga sono the class in all student that sensei o sonkeisita wake]S de wa ai]S teacher respected that is not 'It is not so that all the students in the class respected that teacher.‘ Thus, ssentence (2.65) is still another example demonstrating that negation must occur in a higher sentence to include a quantifier in its scOpe. Ilet us cite another related example: (2.663) sono gakuseitati wa sono daigaku no subeteno the students the college in all sensei o sonkeisi nakatta. teacher respect not-did 'All the teachers in the college, the students did not respect. (= The students did not respect any teacher in the college.)' or 'The students did not respect all the teachers in the college.’ Senunmme (2,66) may be ambiguous, but its primary meaning is t . mn;ln WTiich the quantifier is not included in the s00pe of n . ega351011; in this reading it is synonymous with the fOllOWing : 39 (2.67) sono gakuseitati wa sono daigaku no dono sensei the students the college in any teacher mo sonkeisi nakatta. respect not-did 'The students did not respect any teacher in the college.‘ The sec:ondary meaning of (2.66) is that in which the nega- tive.iricludes the quantifier in its SCOpe, and in that case it is synonymous with (2.68): (2.663) [[sono gakuseitati wa sono daigaku no subeteno the students the college in all sensei o sonkeisita wakeJS de wa nai]S teacher respected that is not 'It is not so that the students respected all the teachers in the college.’ Buttflua secondary meaning of (2.66) is presumably marginal sum.it aseems to have this meaning only when it is supple- Imnmed try the accompanying context as in the following: (2439) sono gakuseitati wa sono daigaku no subeteno the students the college in all sensei o sonkeisi nakatta ga nanninkano sensei teacher respect not-did but some teacher 0 sonkeisita. respected 'The students did not respect all the teachers in the college but respected some of them.’ The acceIYtability of (2.69) indicates that sentence (2.66) ma h y 8N1; ‘the secondary meaning, though marginal, in certain Conte th such as (2.69). The point here is that sentence (2.68) . ‘lfis still another example that involves sentential 40 knerarphy so that negation can include a quantifier in its scope. From the foregoing discussion, it should now be clear tmat semmential hierarchy is used in Japanese so that nega- ‘fion away include in its scope quantifiers that are otherwise notirnzluded in the s00pe of negation. 2.5. gtuantifiers and Contras tive ya _i_n_ J apane s e fifaking the preceding argument one step further, we may gDOn 1:0 discuss sentences such as (2.70) in connection with (2.71) : (2.7CD) sono kurasu no gakusei ga minna Ea sono sensei the class in student all that teacher 0 sonkeisi nakatta. respect not-did 'Not all the students in the class respected that teacher.‘ (23KL) sono kurasu no gakusei ga minna sono sensei o the class in student all that teacher sonkeisi nakatta. respect not-did 'All the students in the class did not respect that teacher.’ Co - Iqmr1n€§ these two sentences, we note that the only dif- fer . ence ls the presence of the contrastive wa in (2.70) and its absence in (2.71). This apparently slight difference CaHSe S a. Esemantic difference between them: (2.70) is Swmn moms with (2.72), while (2.71) is not: 41 (2.72) [[sono kurasu no gakusei ga minna sono sensei the class in student all that teacher 0 sonkeisita wake]S de wa nai]S respected that is not 'It is not so that all the students in the class respected that teacher.’ Itis <3bserved that in (2.72) the negative nai is in a Ingher‘ sentence than the quantifier and, as expected from 'fius sisructure, the former includes the latter in its scope. {Mm syxionymity of (2.70) with (2.72) suggests that the mxmti;fier of (2.70) is also included in the scope of negation. ZIn.turn, sentence (2.71), but not (2.70), has a meaning synonymous with (2.73):9 (2573) sono kurasu no dono gakusei mo sono sensei o the class in any student that teacher sonkeisi nakatta. reSpect not-did 'Lit. *Any student in the class did not respect that teacher. (= No student in the class respected that teacher.)' Hmvcan (nae account for this semantic difference between (25KU aqua (2,71)? Ioreover, how can the synonymity of “570) wigth (2,72) be explained? A natural solution to this Problem is in terms of certain transformations which are by defi ' - niticxn_meaning-preserving. A fuller discussion of this too' - .lc W341;L be given in the following section. 42 2.6. A Proposed Analysis p_f Negation E Japanese Our suggestion concerning the above-observed problem is as follows: (2.74) Sentences (2.70) and (2.72) are derived from the same underlying structure that is significantly different from that underlying sentence (2.71). More specifically, we derive sentences (2.70) and (2.72) from the underlying structure (2.75) in terms of the following transformations whose Operations are shown stage by stage as follows: i. Underlying structure (with unrelated details aside):lo (2.75) ::/S\ 1\me not Quant NP/ V past /\ sOHO kurasu no minna sono sensei sonkeisuru the class in all that teacher respect gakuSei ga 0 Student ii. . . . . . . Application of Sentence-raiSing which raises an emb edded sentence into the immediately higher sentence: 43 (2. 76) LIP/NW Quant l ono rasu minna no sonkeisuru nai no gakusei ga sensei 0 iii. Application of Contrastive lag—attachment which attaches _w_a_ in (2.76) to the quantifier minna "all": (2.77) /3 N:>\\AVP \m /VP\V :4... :L. /\ l l sono kurasuw“ minna wa sono sonkeisuru nai no gakusei ga sensei o iv. Application of Negative-attachment which attaches the negative nai to the verb, after the attachment of 11a to the QMmmifier: (2.78) 31 / \V\ ”P NP/ \ TX Quant past /\ \ ignga rasu minna wa fino sensei Neg 1fusei ga o \ sonkeisuru nai The structure (2.78) becomes sentence (2.70), as proposed here. The generation of sentence (2.72) from (2.75) will be discu Ssed later at the end of this section. 44 The point here is that these transformations must not be gg; hgg; if they are, the above derivation based on these trarmsformations becomes ad hgg as well. The next step is, therwafore, to demonstrate that these transformations are not ad £519 in Japanese. Let us take up Sentence—raising first. Compare the follrywing pair of sentences: (2.'79) a. [watasi ga omou El: sono sensyu wa sonna I think that player such bakana koto o si nai daroo]S foolish thing do not will 'That player, I think, will not do such a foolish thing.‘ b. [watasi HE [sono sensyu ga sonna bakana koto I that player such foolish thing o si nai daroo to]S omou]S do not will that think 'I think that that player will not do such a foolish thing.’ First;’ we observe that sentences (2.79a) and (2.79b) are Synononmous.ll Second, a simple test can show that they have the Slzrface structures shown here: if we insert the comple- mentizer _t_<_)_ "that" in the end of (2.7%) or delete 322 "that" in(23,791)), the resulting sentence is ungrammatical in either case: (2.63C>) a” *watasi ga gmgu,ni, sono sensyu wa sonna I think that player such bakana koto o si nai daroo £9. foolish thing do not will that 45 'Lit. That that player, I think, will not do such a foolish thing.’ b. *watasi HQ sono sensyu ga sonna bakana koto I that player such foolish thing o si nai daroo omou. do not will think 'Lit. That player will not do such a foolish thing, I think.‘ It must be noticed here that in Japanese no NP complement sentence can occur without a complementizer. The ungram- nnticetlgity of (2.80a) and (2.80b), therefore, demonstrates that tliexy have the surface structures shown in (2.79). chnw, considering the synonymity of (2.79a) and (2.79b), W91Mlirrtain.that they are derived from the same underlying structure; (2.8 .NP ”””’/,,VP‘\\\\\\\‘ Aux N' NP V pres watasi ga A2 omou NP ’////2VP\\\\\\‘ Aux sono sensyu NP V Neg daroo that player /\ ‘ ) Will éfihna bakana suru nai 8E1 such foolish do not koto 0 thing This 111'). <3.erlying structure may be given some further support, in ad . d*l*tiion to the above considerations. Recall that the 46 discussion in section 2.2 establishes that the sc0pe of Inegyation.is the whole sentence in which it occurs. On the tnasijs of this constraint, the structure (2.81) indicates that stacea the sc0pe of the negative is 82, watasi gg omou "I thirflc" is outside of its SCOpe. If (2.81) is an acceptable Inuieinlying structure for sentences (2.79a) and (2.79b), theses sentences must be in accord with (2.81) regarding the sc0pea of the negative nai. This is exactly the case here: obserwre the grammaticality of the sentences in (2.82) and the uumgrammaticality of the sentences in (2.83): (2.632) a. watasi ga omou ni, sono sensyu wa sonna I think that player such bakana koto o si nai daroo, sukunakutomo foolish thing do not will at least watasi Kg soo omou. I so think 'That player, I think, will not do such a foolish thing, at least, I think so.’ watasi wa sono sensyu ga sonna bakana koto o I that player such foolish thing si nai daroo to omou, sukunakutomo watasi ya do not will that think at least I SO think 'I think that that player will not do such a foolish thing, at least, I think so.‘ 47 (2 . 83) a. *watasi ga omou ni, sono sensyu wa sonna I think that player such bakana koto o si nai daroo, sukunakutomo foolish thing do not will at least watasi Kg soo omowa nai. I so think not 'That player, I think, will not do such a foolish thing, at least, I don't think so.’ b. *watasi wa sono sensyu ga sonna bakana koto o I that player such foolish thing si nai daroo to omou, sukunakutomo watasi wa do not will that think at least I soo omowa nai. so think not 'I think that that player will not do such a foolish thing, at least, I don't think so.‘ AS menitioned in section 2.2, a constituent, if it is in the sc0pe cxf negation, may be negated by the negation. Thereftxre, if watasi ggygggE El in (2.79a) and Egigél Ea QEQE irl (2.79b) are inside the sc0pe of negation, they can be ne{-Zated and so the sentences in (2.83) should be grammeticai. ( The ungrammaticality of the sentences in 2‘83) and the grammaticality of the sentences in (2.82) damnhitrate that watasi gg omou n; in (2.79a) and watasi E2 0 . ‘QQE 1J1 (2.79b) are outside the s00pe of negation, and this s celfirectly'predicted by the structure (2.81). These Ob SelfiVErtions will be sufficient support for the underlying St r11ctuiee (2.81). -thw, let us consider the derivation of sentences (2.7 Egel) and (2.79b) from (2.81). The derivation of (2.79b) 48 requires the insertion of the complementizer 33 "that" into S2.l2 IIn turn, the derivation of (2.79a) involves the raisinng; of 82 into 81' Applying Sentence-raising after the applix:srbion.of the relevant transformations in the 82—cycle, we derive the intermediate structure (2.84): (2.84) //31\VP\ NP NP VP Aux l l | T V pres Wat asi ga ozlnou sono sensyu sonna bakana koto that player such foolish thing ga o si nai daroo do not will Efince (:2.84) contains no embedded complement sentence, the comPlementizer-insertion rule cannot apply. Moreover, if no furthfi-Z‘I.‘ relevant transformation applies to (2.84), an ‘H@Tamnuatical sentence, (2.85), is derived: (2°853)(=2.80b) *watasi wa sono sensyu ga sonna bakana I that player such foolish koto o si nai daroo omou. thing do not will think 'Lit. That player will not do such a foolish thing, I think.‘ If 'Cnl tile other hand, the adverbial marker g; is attached tot he verb omou "think" in (2.84) and it is moved to preCe d‘3 ‘the subject noun phrase sono sensyu gg "that player," it1u_ . tlmatelyl3 yields sentence (2.7921)- 49 Tflie foregoing discussion indicates that Sentence- raisilig: is necessary to relate the synonymous pair of sentences such as (2.79a) and (2.79b) and similar others such as the following: (2.2363) a. [watasi gg sinzuru gi, Tom wa kessite tomodati I believe ever friend 0 uragira nai]S betray not 'Tom, I believe, never betrays his friends.’ b. [watasi Ea [Tom ga kessite tomodati o uragira I ever friend betray nai tols sinzuruls not that believe 'I believe that Tom never betrays his friends.‘ (2.8V?) an [watasi ga handansuru gi, ano otoko wa saiban I judge that man suit ni makeru daroo]S lose will 'That man, I judge, will lose his suit.’ b. [watasi HQ [ano otoko ga saiban ni makeru I that man suit lose daroo to]S handansuru]S will that judge 'I judge that that man will lose his suit.’ (2‘85?) an [watasi ga soozoosuru gi, zitai wa issoo I imagine situation still akkasuru daroo]S get worse will 'The situation, I imagine, will get still worse.’ 50 [watasi Hg [zitai ga issoo akkasuru daroo I situation still get worse will to]S soozoosuru]S that imagine 'I imagine that the situation will get still worse.‘ It sknatild be observed that in these examples all the verbs belormg 'to the class often called "verbs of saying and fifixflcirng" which are typically characterized in Japanese by thezfaxzt that they take the complementizer :9 "that." But the Sesrrtence-raising transformation is not restricted to this class of examples. Itnother major class involving the application of Sentence—raising contains so-called sentential adverbials. Tb takxe a concrete example, consider the following: (2-89) a. Ift [tasikani nanika ga okottals certainly something happened 'Certainly something happened.’ [[nanika ga okotta koto]S wa tasika da]S something happened that certain is 'It is certain that something happened.‘ ~§§2£§agi "certainly" in (2.89a) is called a sentential Emverbietl, tasika "certain" in (2-89b) may be called a Sen - . . . . . . tent1431 adjective. A Similar test used in connection Wi th (2.'79) can easily prove that sentences (2.89a) and (2. 89k?) liave the surface structures shown here. NOW: ObSerVi Ilag’that sentences (2.89a) and (2.89b) are synonymous, we m ay propose that they be derived from the same 51 underlying structure: (2.90) s:L / \\ NP VP Aux I t I S Adj pres a/’////’ <::\\\\\\\\\ ‘ NP VP Aux tasika I I ( certain N V past nani ka ga okoru s omething happen The validity of this underlying structure is confirmed in a way quite similar to that of (2.81). Now, let us consider the derivation of sentences (2.89a) and (2.89b) from (2.90). If Sentence-raising does not apply to (2.90), the insertion of a complementizer and a COpula into S2 and S 1’ respectively, will generate sentence (2.89b). 0n the other hand, the application of Sentence-raising to (2.90) gives the intermediate structure (231): NP P VP Aux ! l I N Afj pres nanika ga okotta tasika happened Then a-P:plying the same adverbial formation as used in eri - V1113 (2.79a) from (2.84) to attach p_i_ to tasika and 52 nurving it to precede the subject noun phrase, we get sentence (2389a). In Japanese, adverbials usually occur sentence-initially or following the subject noun phrase rather'than.sentence-finally. If this adverbial formation does liot apply to (2.91), the resulting sentence is ungram- matical as follows: (2.592) *nanika ga okotta tasika. something happened certain Clther related examples may be cited in the following: (2-935) 3. [akirakani kare wa wareware o damasita]S clearly he us deceived 'Clearly he deceived us.‘ b. [[kare ga wareware o damasita koto]S wa he us deceived that akiraka da]S clear is 'It is clear that he deceived us.’ (2.921) a. [kanozyo wa matigainaku siken ni gookakusuru]S she no doubt examination pass ’She will undoubtedly pass the examination.‘ b. [[kanozyo ga siken ni gookakusuru nOJS wa she examination pass that matigainai]S no doubt 'Lit. It is no doubt that she will pass the examination. (= There is no doubt that she will pass the examination.)' (2 '95 ) a. [koounnimo kare wa si 0 manugaretals fortunately he death escaped 'Fortunately he escaped death.‘ 53 b. [[kare ga si 0 manugareta no]S wa kooun da]S he death escaped that fortunate is 'It is fortunate that he escaped death.‘ (2.5963) a. [osikumo kare wa annani wakakusite sinda]S regrettably he so young died 'Lit. Regrettably he died so young.‘ b. [[kare ga annani wakakusite sinda hols wa he so young died that osii]n regrettable 'It is regrettable that he died so young.‘ The derrivations of these examples containing sentential adjectives require, as shown above, the application of Sentence-raising as well as Adverbial-formation used in the derivations of (2.79a), (2.8621), (2.878.) and (2.88a): in Imrtieular, they indicate that Sentence—raising in Japanese )3 not restricted to examples containing verbs of saying and thinking. The third major class involving the application of Sentenee-raising includes several classes of adverbials such maadverbials of frequency, nominal adverbials, and adver- tnals of'reason and purpose. A detailed discussion of these adverbials will be given in the next chapter and it will be ‘mmnstreted there that the derivation of sentences 1mrole-ng these adverbials requires the application 01' sentence~raising. Tule foregoing discussion regarding three major classes involvir1 . . . . . g the application of Sentence-raiSing Wlll be 54 sufficxient to show that Sentence-raising is not an gd hoc transformation in Japanese . {Taking the above discussion into consideration, we may give Sentence-raising as: (2.97) OFTEMEXJO N] --)[M X N] o S S 2 l 1 where M, N and X are variables, and M or N dominates VS°T or exhaustively dominates Adjs, Ava, AdvF, AdvR-P or Neg NOticee that Sentence-raising is an optional transformation. Our second and third transformations are Contrastive Xi‘attachment and Negative-attachment. These two transfor- mations are closely related to each other in that Negative—attachment must obligatorily apply after the application of Contrastive Eg-attachment to derive gram- metical sentences. This reflects the fact that all the examples cited in sections 2.5 and 2.5 contain both a Contrastive E3 and a negative. Consequently, it will be more suitable to examine these two transformations together, Iether than separately. To show that these transformations emermt ad hoc in Japanese, let us consider the following: (2.98) a.(=2.53) kanozyo wa koohukuni sini Kg si nakatta. she happily die do not-did 'She did not die happily.‘ b.(=2.55) kanozyo wa koohukuni Kg sina nakatta. she happily die not—did 'She did not die happily.‘ 55 (2.99))(=2.54) [[kanozyo wa koohukuni sinda wake]S de she happily died that is wa nai]S not 'It is not so that she died happily.’ It ShOIlldbe recalled that sentences (2.98a) and (2.98b) are synonymous with (2.99), as discussed in section 2.5. Thus, bvml (2398) and (2.99) may be assumed to derive from the Same linderlying structure: NP VP a I ( (2.100) NP V Aux nai l / \ not ? Ade V past kanozyo ga koohukuni Sinu she happily die The derivation of (2.99) from (2.100) involves the insertion ofacomplementizer and a copula into 82 and 81’ reSpectively. Inturn, the derivation of (2.98) from (2.100) requires the musing of 82 into 81' Next, the contrastive Kg is attached tothe verb as in (2.101) or to the adverbial as in (2.102): (2 101) /S)l\,\ NP VP ux VP I ////’ \\\\\ I ‘ N Adv,VI V past Neg I I ‘ / kanozyo ga koohukuni sinu wa nai (2.102) NP/S\'IVP\\VP \Aux N Ade past Neg I I kanozyo ga koohukuni wa sinu nai Then, the Negative-attachment rule applies to attach the Negative to the verb, yielding sentences (2.98a) and (2-98b), respectively. The application of Negative— attachment is obligatory, since if this transformation is not applied, the resulting sentences are ungrammatical as fOllows: (2.103) a. *kanozyo wa koohukuni sini wa sita nai. she happily die did not b. *kanozyo wa koohukuni wa sinda nai. she happily died not MOreover, notice that when a contrastive Kg or a negative is attached to a verb, it causes the following morphological Changes: (2.104) a. sinu+wa past -9 sini wa sita die die did cf. sinu past -—) sinda die died 57 b. sinu+nai past -9 sipg nakatta die not die not-past cf. sinu past nai -—9 *sinda nai die not died not c. sinu+wa+nai past--9 sip; wa si nakatta die not die do not-past cf. sinu+wa past nai -—9‘*sini wa sita nai die not die did not The Hubrphological changes indicate that fig and a negative are ruat just shifted to follow a verb but are attached to a verb. From the above derivation it is clear that Negative- attachment in Japanese is closely related to Contrastive Ha-attachment, namely, it can apply only after Contrastive Ea-attachment has been applied. Furthermore, examples such as (2.98) indicate that Contrastive Eg-attachment and Negative-attachment are not restricted to the derivation of Sentences containing a quantifier and negation. Ineidentally, notice that the derivation of (2.98) from ULIOO) also involves the application of Sentence-raising, Thus providing additional support for this transformation. In connection with Contrastive Eg-attachment and Negative-attachment, we may further cite the following: (2.105)(=2.45) [kore wa mattaku tadasiku.yg nai]S this altogether correct not 'This is not altogether correct.’ 58 (£2.I106)(=2.46) [[kore wa mattaku tadasii wake]S de this altogether correct that is Kg nails not 'It is not so that this is altogether correct.‘ As tuited in section 2.3, sentence (2.105) is synonymous with (2.1136). Observe that neither (2.l05) nor (2.106) contains a qiuantifier but the adverbial mattaku "altogether" instead. If'tne adopt the above analysis in terms of Contrastive Eg- attacMent and Negative-attachment as well as ESentence-raising, the synonymity of sentences (2.105) and (2.106) involving no quantifier can be explained in a way quite similar to that of (2.70) and (2.72) involving a quantifier. Under this analysis, sentences (2.105) and (2.106) will be derived from the same underlying structure in the following way: i. Underlying structure: (2.107) / Sl\\ TP Vf wa q , ////’SZ::::\\\“~1‘~11~ 1T5 HP VP Aux nai i / \ . ‘ I101: T A?VD ij pres kore ga mattaku tadasii this altogether correct 59 ii. Application of Sentence—raising: (2.108) 81 MN NP ////VP ATx Vf wa N Advb ‘\\\\Adj pres Neg kore ga mattaku tadasii nai iii. Application of Contrastive Ea-attachment: NP /////y£\\\\\\ ATX VP N AdvD Adj pres Neg .korl ga mattaku tadasii wa nai iv. Iipplication of Negative-attachment: (2.110) //sl\ NP VP Aux N Ad””’/’/i‘\\\\\ldj pies v D | I \Ng kOre ga ma taku T tadasii wa nai (Hm Structure (2.110) yields sentence (2.105). The relevant difference between (2.70) and (2.105) is that the contra-Stive [a is attached to the quantifier in (2.70), While in (2.105) it is attached to the adjective modified by the adverbial, not to the adverbial itself. But this dif- .fir enc“3 dissolves when Contrastive Eg-attachment applies to 60 (2.108) in such a way as to attach ya to the adverbial m "altogether," giving (2.111) instead of (2.109): (2.111) s:L /I \\V M) VP Aux P I / \ . I I N Adv Adj pres Neg kore ga mattaku wa tadasii nai The structure (2.111) gives sentence (2.112), which is synonymous with (2.105): (2.112) kore wa mattaku 1g tadasiku pal. this altogether correct not 'This is not altogether correct.’ It should be noted that in (2.112) the contrastive t_v_a_ is attached immediately to the adverbial, just as Kg is attached to the quantifier in (2.70)- 3Turning to the derivation of sentence (2.106) from (2.UT]), the insertion of a complementizer and a copula into 822nm.E31, respectively, gives sentence (2.106). 3From.the above discussion it will be clear that Cmmrastive wgeattachment and Negative—attachment are inde- pendently motivated transformations in that they are necessar'yto account for the synonymity of (2.98) with (2°99) and that of (2.105) and (2.112) with (2.106). Taking into consideration the foregoing discussion, Co ntrafirtive Eg—attachment will be given informally as (fill3) ‘ 61 (2.];L3) Contrastive Eg-attachment in Japanese attaches a contrastive ya to a quantifier, a manner adver- bial, a verb modified by a manner adverbial, a degree adverbial or an adjective modified by a degree adverbial. This (man be formalized into: (2.114) Quant Ade X [Adv Adv D [AdvD . . . AdJIVP V] Y wa Z -9 M ' ° ° VP Quant+wa Ade+wa X [Ade . . . V+wa]VP Y Z AdvD+wa [AdvD . . . Adj+wa]VP where X, Y and Z are variables {Turning next to Negative-attachment, it will be given informally as (2.115), to cover the derivation of (2.98a) and (2J98b) from (2.101) and (2.102), respectively, and that of (2-105) and (2.112) from (2.109) and (2.111), reSPeetively: (2.115) Negative-attachment in Japanese attaches a negative to the verb or the adjective that is combined with a contrastive Kg or is preceded by a constituent combined with a contrastive Kg. This wi£11 be formalized as: (2.116) V } P Y {Adj + wa K V Z Neg —-9 Q + wa YIAde P Y (Adj }+ wa + Neg X Z Q+wa Y{Adj]+Neg where P, Q, X, Y and Z are variables, Z contains no Neg, P includes Ade or AdvD and 0 includes Quant, Ade or AdvD Observe that (2.116) requires the presence of a constituent Combined with a contrastive pg, reflecting the fact that IIegative-attachment can apply only after the application of Contrastive Eg-attachment. Furthermore, Negative-attachment is not a rule that inserts some element in a higher sentence into a lower embedded sentence. Notice, in this regard, that the application of Sentence-raising, preceding that of Ikgative-attachment, has moved the verb to be combined With anegative up into the sentence in which the negative occurs. The above arguments will be sufficient to demonstrate Hum Contrastive Ea—attachment and Negative-attachment, as well as Sentence—raising, are, in any case, necessary in Japanese grammar. In particular, examples such as (2 98) (2.105) and (2.112) are extremely significant, since they serve to demonstrate the independent motivation of not only Contrastive Eg-attachment and Negative-attachment but also Sentence-raising. Other similar examples may be cited 63 in the following: (2.ll7)(=2.47) [sono siyoonin wa taihen syooziki de Kg the employee very honest is 14 nails not 'The employee is not very honest.‘ (2.118)(=2.48) [Esono siyoonin wa taihen syooziki na the employee very honest is wake]S de Kg nails that is not 'It is not so that the employee is very honest.‘ Both (2.117) and (2.118) will be derived from (2.119) under our analysis: (2.119) /sl\\ NP ' VP wa I I NP’//// é:::::::::::::VP‘~‘~“P‘Aux nLi Det/ \( Adv/ \c‘ij pllt‘es net I I I” sono siyoonin ga taihen syooziki the employee very honest Another example is: (2.120) a.(=2.49) [sono siyoonin wa kimi yori the employee you more-than syooziki de Kg nai]S honest is not 'The employee is not more honest than you.’ 64 b.(=2.51) [sono siyoonin wa kimi yori Kg the employee you more-than syooziki de nai]S honest is not 'The employee is not more honest than you.‘ (2.121)(=2.50) [[sono siyoonin wa kimi yori syooziki the employee you more-than honest na wake]S de Kg nai]S is that is not 'It is not so that the employee is more honest than you.‘ Sentences (2.120) and (2.121) will be derived from the same underlying structure, (2.122) with unnecessary details aside: (2.122) Flf”/’/,,,»S f::::::::::::::::‘§~“~a NPS/ V... I: De/NP \ /\ , Adv Adj pres DI I A sono siyoonin ga kimi yori syooziki the employee you more- honest than Now, it is time to consider the derivation of sentence (2.72) from the underlying structure (2.75). They will be repeated here for convenience: 65 (2.l23)(=2.72) [[sono kurasu no gakusei ga minna sono the class in student all that sensei o sonkeisita wake]S de wa nai]S teacher respected that is not 'It is not so that all the students in the class respected that teacher.‘ (2.124><=2.75) mps/ \Wwa NP m////YP A nLi ‘\\\\V IX not Quant p st \ sono kurasu no* minna sono sensei' sonieisuru the class in all that teacher respect gakusei ga 0 student When Sentence-raising does not apply to (2.124), the complementizer wage "that" is inserted into 82. Then, the contrastive ya in S1 cannot be attached to a constituent in S because of a universal condition stated by Chomsky 2 (1965, 146), namely, that no transformation can insert mor- phological material into lower sentences. Thus, this we remains unchanged. Next, the insertion of a COpula will derive the following structure: (2.125) [[sono kurasu no gakusei ga minna sono sensei o the class in student all that teacher sonkeisita wake]S da [[naiJN egJVP wa]S respected that 2 is not 81 66 Then, attaching Eé to the COpula Q3 "is,n15 we get sentence (2.123). It should be mentioned here that the underlying structure (2.124) is based on the assumption given in G. Lakoff (1965), Fillmore (1968) and others that a COpula is not present in underlying structure but is transforma— tionally inserted as a kind of tense-carrier. In this connection, we may point out one linguistic fact observed in Japanese: a copula does not appear with ordinary predicate adjectives.l6 To illustrate the point, let us consider the following: (2.126) a. kareno ie wa ookii. his house big 'His house is big.’ a .*kareno ie wa ookii da. his house big is 'His house is big.’ b. ano huzin wa utukusii. that lady beautiful 'That lady is beautiful.‘ b .*ano huzin wa utukusii da. that lady beautiful is 'That lady is beautiful.’ Quite similarly, the negative nai may function as a predicate adjective in Japanese, as in: (2.127) a. kono atari ni wa mizu ga nai. this neighborhood in water not 'Lit. Water is not in this neighborhood. (= There is no water in this neighborhood.)' 67 a .*kono atari ni mizu ga nai da. this neighborhood in water not is 'Lit. Water is not in this neighborhood. (= There is no water in this neighborhood.)' Furthermore, we may mention that in traditional Japanese gnammar the negative nai is included in the same class as ordinary adjectives on the basis of their similar inflec- tional behaviors. These considerations together provide some support for the underlying structure (2.124) in which the copula da "is" is not present. In order to derive underlying structures such as (2.124) we need the following base rules: (2.128) a. s —-) NP VP (Aux) (wa) b. VP -—-) Neg e. V}? —-) (NP) V (Neg) We may call the negation in (2.124) sentential negation which is derived by the base rule (2.128b), as Opposed to verb-phrase negation that is derived by the rule (2.1280). The relevant difference between them is that sentential negation is exhaustively dominated by a verb phrase, commanding the sentence it negates in underlying structure, whereas verb-phrase negation is an Optional constituent of a verb phrase. To summarize, we have demonstrated that Sentence— raising, Contrastive wa-attachment and Negative-attachment are all independently motivated transformations in Japanese 68 {gammar and that they may relate in a general way sentential hierarchy to a contrastive Ea with respect to negation. fibreover, we have distinguished two types of negation, sentential and verb-phrase, in a rather formal way: the former type is derived by the rule (2.l28b), while the latter is derived by the rule (2.1280). The necessity and the motivation for this twofold distinction of negation will be further discussed later in connection with the scope of negation and quantifiers. 2.7. The Contrastive HQ Versus the Topic 1g Attached £2 Quan: tifiers Negative-attachment given in (2.116) indicates that it requires the preceding application of Contrastive Eg- attachment. As regards Contrastive wa—attachment, when adverbials such as koohukuni "happily" and mattaku "altogether" are involved, the contrastive Ea may be attached not only to the adverbials but also to a verb modified by koohukuni or an adjective modified by mattaku. With quantifiers, however, the contrastive ya can be attached only to the quantifier itself, not to a noun modified by the quantifier. What causes this discrepancy in the case of quantifiers? To start with, compare the following sentences: (2.129) John wa ookuno syoosetu HQ yoma nakatta. many novel read not-did 'Many novels, John did not read.’ or 69 'John did not read many novels.‘ (2.130) John wa syoosetu o ooku wg yoma nakatta. novel many read not—did 'John did not read many novels.’ (2.131) [[John wa ookuno syoosetu o yonda wake]S de many novel read that is wa nai]S not 'It is not so that John read many novels.‘ Observe the morphological difference between the quantifier ookung in (2.129) and gggu in (2.130). In (2.129) the quantifier is used attributively, preceding the noun, while in (2.130) it follows the noun, and so it takes the form ooku, not ookuno; in Japanese quantifiers and numerals drop ng when they follow the nouns which they modify. Now, we observe that sentence (2.130) is unambiguous and has the meaning of (2.131). In contrast, sentence (2.129) may be ambiguous with two readings. The primary reading is that many novels, John did not read, with the second Kg being interpreted as a topic 33, combined with the noun phrase ookuno syoosetu "many novels" as a unit. The secondary reading is that of (2.131): in this reading the same as is taken to be a contrastive Kg, semantically associated with the quantifier, though it is structurally attached to the noun modified by the quantifier. One relevant factor that makes the primary reading, "Many novels, John did not read," is that a topic HE is more common and general than a contrastive Ea. In this connection, consider the following: 0“ .‘-l‘ ‘ -. ., ... ... .~-.‘ - ». ... 70 (2.132) John Ea syoosetu o yonde iru. novel reading is -'As for John, he is reading a novel.’ or 'John (in contrast with someone else) is reading a novel.‘ Sentence (2.132) may be ambiguous, as noted in section 2.3, but the primary reading is that as for John, he is reading a novel, with the wg being interpreted as a tOpic wa rather than a contrastive flé- Only when it is accompanied by the context which requires a contrastive ya interpretation may sentence (2.132) have the reading of "John (in contrast with someone else) is reading a novel," as in the following: (2.133) John wa syoosetu o yonde iru ga Mary Ea terebi novel reading is but television 0 mite iru. watching is 'John is reading a novel, but Mary is watching television.‘ The primacy of a tOpic HQ over a contrastive HQ may be further observed in so-called tOpicless sentences. For instance, consider the following: (2.134) ginkoo e ike nakatta yo. bank to can-go not-past F(ina1)—P(article) 'I could not go to the bank.’ (2.135) syukudai o yattyatta kai. homework finished Q(estion)-M(arker) 'Did you finish the homework?’ Colloquial speech often deletes the tOpic of a sentence, especially when the topic is the speaker as in (2.134) or 71 the hearer as in (2.135). If we add the topic to (2.134) and (2.135), the resulting sentences will be (2.136) and (2.137), respectively: (2.136) boku ya ginkoo e ike nakatta yo. I bank to can-go not—past FP 'I could not go to the bank.‘ (2.137) kimi Ea syukudai o yattyatta kai. you homework finished QM 'Did you finish the homework?‘ It Should be noticed that the as in (2.136) and (2.137) can rmver be taken as a contrastive Ea, if these sentences are to be synonymous with (2.134) and (2.135), respectively. This fact suggests that a topic Kg may be deleted without causing any ambiguity and may be recovered, causing no semantic change, in certain contexts, whereas a contrastive ya can never be so deleted or recovered. This difference indicates the fact that a topic Ea is more general or "unmarked" than a contrastive Ea. Furthermore, a topic Ea may occur more than once in a simplex sentence, while a contrastive wa may not. To take a concrete example, consider the following: (2.138) kinoo ya John Ea tosyokan e E2 ika nakatta. yesterday library to go not—did 'John did not go to the library yesterday.’ It is possible to interprete at least two wa's in (2.138) as a tOpic HQ: but it rarely, if ever, happens that the two HE'S out of the three in sentences such as (2.138) may be taken to be a contrastive Eé- These facts will be enough to 72 demonstrate the primacy of a topic Kg over a contrastive Ea. Now, consider the following: (2.l39) subeteno oobosya fig sintaikensa o uke nakatta. all applicant physical- undergo not-did examination 'All the applicants did not undergo a physical-examination.‘ ‘ (2.140) oobosya ga subete Hg sintaikensa o uke nakatta. applicant all physical- undergo not-did examination 'Not all the applicants underwent a physical-examination.’ (2.141) [[subeteno oobosya ga sintaikensa o uketa all applicant physical- underwent examination wake]S de wa nai]S that is not 'It is not so that all the applicants underwent a physical—examination.’ If sentence (2.139) is acceptable, it is not synonymous with (2.141), though sentence (2.140) is synonymous with (2.141). This is quite natural in the light of the primacy of a topic Ea over a contrastive ya: ya in (2.139), attached to the noun modified by a quantifier, is most naturally interpreted as a topic Ea, though wg in (2.140), attached immediately to the quantifier, is a contrastive wg. From the primacy of a topic 1% it naturally follows that if there is only one HQ attached to a noun phrase in a sentence, this Ea is usually taken as a topic E2: and the noun phrase with this 3g is interpreted as the tOpic of a sentence. But when there are two noun phrases with wa, one 73 of them may be interpreted as a contrastive wa. From this viewpoint, reexamine sentence (2.129). repeated here as i (2.142): (2.142)(=2.129) John Ea ookuno syoosetu Ea yoma nakatta. many novel read not—did 'Many novels, John did not read.‘ or 'John did not read many novels.‘ As noted above, sentence (2.142) may be ambiguous and its Inimary reading is that in which the second Ea attached to the noun phrase containing a quantifier is semantically interpreted as a topic Ea as well as the first ya, and its secondary reading is that in which the same Ea is interpreted as a contrastive Eé- The latter interpretation is possible, though secondary, since sentence (2.142) has another noun phrase with HQ: Jghn ya, which is most naturally taken to be the topic of the sentence: the presence of this constituent makes it possible for the second ya to be interpreted as a contrastive Ea. Furthermore, a quantifier following a noun such as subete ya in (2.140), though combined with ya, is a constitu- ent that cannot become the tOpic of a sentence. The same is true of adverbials, adjective phrases and verb phrases which are combined with ya. In contrast, noun phrases combined with ya such as subeteno oobosya ya in (2.139) become most naturally the topic of a sentence. In sum, the above-observed discrepancy--that a contrastive wa can be attached directly to a quantifier 74 following a noun, but not to a noun modified by a quantifier--may be accounted for by the fact that: the Inimacy of a topic 3; over a contrastive Hg makes a noun combined with ya most naturally the tOpic of a sentence; [ M x N 2 l 1 where M, N and X are variables, and M or 19 1S N dominates VS-T or exhaustively dominates Adjs, Adv Adv Adv . N’ F’ R P or Neg The applicability of the same transformation to both Japanese and English gives further support for this transformation. Now, we may go on to consider the second transforma- tion, Negative-attachment. This transformation is roughly similar to what Klima (1964) calls Negative—incorporation, in the sense that a negative is attached to a nominal quantifier when the nominal is subject. Klima (1964, 272) gives Negative-incorporation as follows: (2.181) , §g .111 -—-> [QEEEI‘Noun"(EII}JN6minal-Q25_Tense {hgge A [neg+Quant-Noun- {% } J M Nominal-Tense { be } X have 87 But it is important to note that his Negative—incorporation rule is not meaning-preserving. For instance, his Negative— incorporation rule applies to the underlying structure of sentence (2.182), giving sentence (2.183), which is not synonymous with (2.182): (2.182) Nagy people did gap like the movie. (2.183) N93 aaay peOple liked the movie. In contrast, our Negative-attachment rule becomes meaning- preserving, if we impose on it one constraint to the effect that a negative is attached to the leftmost pre-verbal quantifier, namely, the leftmost quantifier preceding the verb, and if there is no such quantifier, the negative is attached to the verb. Thus our Negative-attachment rule is informally something like (2.184): (2.184) Negative-attachment in English attaches a negative to the leftmost pre-verbal quantifier of, if there is no such quantifier, to the verb which follows no quantifier or negative. This may be formalized as: (2.185) a. x Quant Y Neg——) X Neg+Quant Y b. X V Y Neg -9 X Neg+V Y where X and Y are variables, and X contains no Quant or Neg According to this Negative-attachment rule, the following derivation, for instance, takes place: 88 (2.186) a. [maay arrows past hit the target [QQIJNngS -—§ N91 may arrows hit the target. b. [the police past arrest maay demonstrators [notJNngS ——9 The police did gap arrest aaay demonstrators. Moreover, the Negative-attachment rule correctly attaches £93 to aay as in (2.l87a), but not to the verb as in (2.187b): (2.187) a. [aay girl pres enjoy dating Bill [aathngS —->N_o_t any girl enjoys dating Bill. 0-9.No girl enjoys dating Bill.) b. [aay girl pres enjoy dating Bill [QQLJNngS --9‘EAQ1 girl does a9; enjoy dating Bill. We note that the derivation (2.187b) gives an ungrammatical sentence and this is due to the violation of the condition of Negative-attachment (2.l85b). Thus, the derivation (2.187b) is blocked through the filtering function of Negative—attachment. To show the independent motivation of this transfor- mation, consider the following sentences involving no quantifier but certain kinds of adverbials and negation: (2.188) a. It is pg: so that aeg that boy stole apples in the orchard. b. Nag gaIy that boy stole apples in the orchard. c. QaIy that boy did E22 steal apples in the orchard. 89 (2.189) a. It is gag so that that player aIaag does his best. b. That player does not often do his best. c. That player gigag does gag do his best. We notice that sentence (c) is not synonymous with either sentence (a) or sentence (b) in both (2.188) and (2.189). This fact suggests that the negative gag in the (b) sentences must be distinguished from that in the (0) sentences in some way. It may be prOposed here that these two types of negation be distinguished in terms of the dis- tinction between sentential and verb—phrase negation: gap in the (a) and (b) sentences is sentential negation which is derived by the base rule (2.l90b), while gag in the (0) sentences is verb-phrase negation that is derived by the rule (2.1900): (2.190) a. s —-)NP (Aux) VP b. VP ---9 Neg c. VP —-9 (Neg) V (NP) This distinction between sentential and verb-phrase negation is different in a significant way from the dis- tinction between sentence and constituent negation in Klima (1964), and that between sentence and verb phrase negation in Jackendoff (1969). According to Klima, the negation in both (2.188) and (2.189) is sentence negation. In Jackendoff's analysis, on the other hand, gag in [ X - neg - Y ]S is sentence negation if there exists a paraphrase "It is not so that [X - Y IS." Thus, the 90 negation in (2.1888), (2.188b), (2.1898) and (2.189b) will be sentence negation, while that in (2.1880) and (2.1890) will be verb phrase negation. Furthermore, according to his analysis, the negation in (2.1918) and (2.1928) will be sentence negation, since there are paraphrases of (2.1918) and (2.1928), such as (2.l91b) and (2.l92b), reSpectively: (2.191) 8. That boy did gag steal apples in the orchard. b. It is gag so that that boy stole apples in the orchard. (2.192) 8. That player does gag do his best. b. It is gag so that that player does his best. In our analysis, however, the negation in (2.1918) and (2.1928) is verb-phrase negation, whereas that in (2.l91b) and (2.l92b) is sentential negation. These differences may be summarized as: (2.193) Klima Jackendoff Our Analysis Sentential Sentence Neg. Sentence Neg. Neg. Verb-phrase Verb Phrase Neg' Neg. Thus, Klima's sentence negation includes Jackendoff's verb phrase negation and Jackendoff's sentence negation includes part of our verb-phrase negation. In addition, our sentential negation is different from both of their sentence negation in that it is generated outside the sentence which 91 it negates in underlying structure, while their sentence negation is derived inside the sentence which it negates. Now, let us return to the sentences in (2.188) and (2.189). Under our analysis, sentences (2.1888) and (2.188b) are derived from the underlying structure (2.194), whereas sentences (2.1898) and (2.l89b) are derived from (2.195) with unnecessary details aside: TP/ ////§2\ NLg Aux V//’ not past P\\:(P Z/////9\\\\\L only that boy steal in the orchard apples .// N:\ T \\dv VF VP not Al I / \ pres often V NP that player do his best The structure (2.194) indicates that since the scope of not is 81’ only is included in its 800pe. This accounts for the fact that in (2.188a) and (2.188b) only is included in the 92 scope of gag, as opposed to (2.1880) in which aeg is outside the scope of BEE-20 Similarly, the structure (2.195) indicates that aIgag is in the SCOpe of gap, reflecting the fact that in (2.1898) and (2.189b) aifiag is included in the scope of gag, unlike (2.1890) in which gigag is outside the 800pe of gag?l Now, considering the derivation of (2.188b) from (2.194) and that of (2.189b) from (2.195), they require the attaching of gag to aeg and gigag, respectively, as well as the raising of 82 into S1. The point here is that this attaching of gag can be taken care of by the Negative- attachment rule in (2.185), if it is slightly revised to: (2.196) a. Quant X { Ava Y Neg "9 AdvF Quant X N eg+ { Ava Y AdvF b.(=2.185b) X V Y Neg'-$ X N e g+V Y where X and Y are variables, and X contains no Neg, Quant, AdvF or Ava such as aeg The fact that Negative-attachment is necessary to account for the synonymity of (2.188b) with (2.1888), and (2.l89b) with (2.1898) demonstrates that this transformation is not restricted to examples involving quantifiers and negation. Moreover, examples such as (2.188b) and (2.189b) further serve to confirm the independent motivation of 93 Sentence-raising. The remarkable difference between Negative-attachment in Japanese and that in English is that a negative is always attached to a verb (including an adjective and a copula) in Japanese, while in English it is attached primarily to a quantifier, then to a verb if there is no quantifier preceding the verb. But it should be recalled that in Japanese 8 contrastive ga is attached to 8 quantifier before the application of Negative—attachment. Thus, the following parallel is observed between them: (2.197) a. Japanese: i. [ . . . Quant . . . waJS'-—9 [ . . . Quant+wa . . . IS ii. [ . . . Quant+wa . . . V . . . Neg]S -—$' [ . . . Quant+wa . . . V+Neg . . . JS [ . . . Quant . . .‘V . . . Neg]S -—fi> [ . . . Neg+Quant . . . V . . . JS In Japanese, only after ga is attached to a quantifier as in (2.l97a.i) can a negative be attached to a verb preceded by a quantifier as in (2.l97a.ii). In English, on the other hand, a negative must be attached to a pre-verbal quantifier as in (2.l97b); a negative cannot be attached to a verb preceded by a quantifier. This restriction in English may be compared with the restriction in Japanese that ga must be attached beforehand to a quantifier so that a negative can be attached to 8 verb preceded by a quantifier. 94 It is now time to turn to the underlying structure (2.160), which will be repeated here for convenience: NT//////Slf\\\\\\“\\“VP l | 1/1:\ Ii \ 1 V/ \1. Qufnt N past (2.l98)(=2.l60) many arrows hit the target This structure will be generated by the above base rule (2.190). In particular, it is important to note that nega- tion in (2.198) is derived by the rule (2.l90b), not the rule (2.1900). It should also be noticed that the underlying structure (2.198) does not contain I: preceding the embedded sentence: we assume, following Fillmore (1968) and Langendoen (1969), that the expletive I: does not exist in underlying structure, though Rosenbaum's (1967) argument is based on its presence in deep structure. In this connection, one linguistic fact may be mentioned: in Japanese there is no equivalent of this English I: at 811.22 Moreover, the structure (2.198) involves the assumption given in G. Lakoff (1965), Fillmore (1968) and others that a 00pula is not present in underlying structure but is transformationally inserted as a kind of tense-carrier. In this connection, it may be mentioned that in Japanese 8 copula does not appear with ordinary predicate adjectives, 95 as discussed in section 2.6. Now, let us consider the derivation of sentence (2.1598), repeated here as (2.199) for convenience: (2.l99)(=2.l598) [It is gag so [that gagy arrows hit the targetJS]S The derivation of (2.199) from (2.198) must involve the insertion of IE, Ia and aa as well as the extraposition of the embedded sentence (as in Rosenbaum (1967)). Here, it should be recalled that the parallel derivation of the Japanese correspondent of (2.199), namely, (2.l59b), must also involve the insertion of the 00pula ga "is," as discussed in section 2.6, though the equivalents of I; and aa are not inserted: there is no equivalent of IE at all in Japanese, as mentioned in footnote 1?, and the insertion of ag does not seem to be obligatory even in the derivation of English (2.199) and other similar sentences.23 Summarizing so far, we have argued that sentences such as (2.1578) and (2.1598) are derived from the same underlying structure, and that the transformations neces- sary for the derivation of sentences like (2.157a) are independently motivated in English grammar. This analysis, if adOpted, may solve some of the problems noted in section 1.2 of Chapter I, namely, problems A through D. Let us examine each: A. According to our analysis, sentences (1.5) and (1.6), repeated here as (2.200) and (2.201), will be derived from the underlying structures (2.202) and (2.203), 96 reSpectively: (2.200)(=l.5) Not many arrows hit the target. (2.201)(=l.6) Many arrows did not hit the target. (2.202) [[[Emany arrows]NP [past]Aux [[hit]V [the targethPJVPJSZJNP [EggplNeglVPISl (2.203) [[magy arrows]NP [past]Aux [[gailNeg [hit]V [the targetJNPJVPIS It is important to note that the negation in (2.202) is derived by the base rule (2.l90b), while that in (2.203) is derived by the rule (2.1900). Thus, the semantic dif- ference between (2.200) and (2.201) can be explained by the relevant difference of their underlying structures, in particular, the sentential negation in (2.202) versus the verb-phrase negation in (2.203). Similarly, sentences (1.8), (1.9) and (1.10), repeated here as (2.204), (2.205) and (2.206), will have the underlying structures (2.207), (2.208) and (2.209), respectively, under our analysis: (2.204)(=l.8) The police did gay arrest gagy demonstrators. (2.205)(=1.9) Nagy demonstrators were gay arrested by the police. (2.206)(=l.10) Nay gagy demonstrators were arrested by the police. (2.207) [[[[the police]NP [pastJAuX [[arrestJV [gagy demonstratorslNPJVPJS 1 2 NP [EEQEJNngVPJS 1 97 (2.208) [[the police]NP [past]Aux [[ggtheg [arrest]v [many demonstratorslNPJVPJS (2.209)(=2.207) [[[[the policeJNP [past]Aux [[arrestlv [many demonstratorslNPJVP]SZJNP ”MJNngVPJSl We note that the structure underlying sentence (2.204) is the same as that underlying sentence (2.206), but is dif- ferent from that underlying sentence (2.205). This provides an explanation for the fact that (2.204) is synonymous with (2.206), but not with (2.205). B. The counter-example of the first type may be accounted for within our framework. For instance, sentence (1.11), repeated here as (2.210), will have the underlying structure (2.211) in our analysis: (2.210)(=l.11) The arrows that did gag hit the target were gagy. (2.211) [[[the arrows]NP [arrows past not hit the targetJSZJNP [past]Aux [[manyJQuantJVPJSl In (2.211) the quantifier gagy in the higher sentence includes the negative gay in its SCOpe. Thus, sentence (2.210) is synonymous with sentence (1.6), repeated above as (2.201), in which gagy includes the negative in its scope. The second type of counter-example, in which the quantifier in question has extra heavy stress, cannot be explained in our analysis given so far. 0. This analysis does not involve the application of 98 Quantifier-lowering, but rather Sentence-raising and other transformations which are all independently motivated, as demonstrated above. D. Our underlying structures for sentences with negation and quantifiers are not as ag gag as the generative semanticians' (e.g. G. Lakoff's), since the underlying structures and their corresponding surface structures are related by means of transformations independently moti- vated in both Japanese and English. But there are still a few problems which remain unsolved in this analysis. To start with, consider and compare the underlying structures (2.202) and (2.203) once more. The proposed analysis cannot block the application of Passivization to (2.203) underlying (2.201) to yield sentence (1.7), repeated here as (2.212), in spite of the fact that (2.212) is not synonymous with (2.201): (2.212)(=l.7) The target was gay hit by gagy arrows. Similarly, this analysis is not satisfactory in that when Passivization does not apply to (2.208) underlying (2.205), the resulting sentence is (2.204), though (2.204) is not synonymous with (2.205); only when Passivization applies to (2.208) will sentence (2.205) be generated, as desired. These considerations lead us to the conclusion that (2.203) is not an acceptable underlying structure for (2.201), and neither is (2.208) for (2.205). Thus, the prOposed analysis is not complete by itself but needs to be 99 supplemented by 8 fuller analysis of quantifiers. 2.9. a Proposed Analysis af Quantifiers with Respect ya Negation Since the proposed analysis of negation needs to be supplemented by a fuller analysis of quantifiers, we may now proceed to consider and prOpose an analysis of quantifiers with respect to negation. The analysis of quantifiers to be proposed will be based on the similarities between quanti- fiers and negatives.24 2.9.1. a Proposed Analysis a: Quantifiega,Ig Japanese In discussing negation with reSpect to quantifiers in section 2.5, we have mentioned that gIgga "all" in sentence (2.214) is included in the s00pe of negation, as contrasted with gIgga "all" in (2.213) which is not included in the scope of negation: (2.213)(=2.71) sono kurasu no gakusei g8 minna sono the class in student all that sensei o sonkeisi nakatta. teacher respect not-did ’All the students in the class did not respect that teacher.‘ (2.214)(=2.70) sono kurasu no gakusei g8 minna ya sono the class in student all that sensei o sonkeisi nakatta. teacher respect not-did 'Not all the students in the class respected that teacher.‘ 100 If our analysis of negation is acceptable (in particular, if the negation in (2.214) has originated from a higher sentence), then the logical extension of this analysis permits us to derive gIgga "all" in (2.213) from a higher sentence. This analysis of negation results from the observation that the SCOpe of negation is the whole sentence in which it occurs, as noted in (2.18). It follows from this that if a constituent is not included in the 800pe of negation, it is outside the sentence containing the negation. Therefore, if sentence (2.213) is not synonymous with (2.215) involving the negation in a higher sentence which includes the quantifier in its s00pe, it is because gIgga "all" in (2.213) is not included in the scope of negation: (2.215)(=2.72) [[sono kurasu no gakusei ga minna sono the class in student all that sensei o sonkeisita wake]S de wa nai]S teacher respected that is not 'It is not so that all the students in the class respected that teacher.’ This is clearly confirmed by the fact that (2.213) may be synonymous with (2.216), in which the quantifier subeteno "all" is outside the sentence containing the negative: (2.216) [[sono sensei o sonkeisi nakatta no]S wa sono that teacher respect not-did that the kurasu no subeteno gakusei datta]S class in all student was 'Lit. It was all the students in the class that did not respect that teacher.‘ 101 In (2.216) the quantifier, which occurs in a higher sentence than the negative, is outside the s00pe of the negative. Now, the synonymity of (2.213) with (2.216) along with the non—synonymity of (2.213) with (2.215) suggests that the quantifier gIgga "all" in (2.213) occurs in the underlying structure not only outside the sentence containing the negation but also in a sentence higher than the negation. From the foregoing argument it follows that if our analysis of negation--based on the constraint that the scope of negation is the whole sentence in which it occurs--is acceptable, the underlying structure for sentence (2.213) will be (2.217) with unnecessary details aside: :i’/////”S17777557777‘IIIIIP“VP minnano NP\\\\N I>\\\\SNE811 NP past sono gakusei sonkei- nai suru the student sono not that respect (2.217) kurasu no ga class in sensei 0 teacher This underlying structure indicates that since the s00pe of the negative is 82, the quantifier is outside of its scope. The quantifier minnano "all" may be called a sentential quantifier, as opposed to a nominal quantifier such as 102 gIgga ”all" in (2.214). A sentential quantifier is derived by the rule (2.218b), while a nominal quantifier is derived by the rule (2.2180): (2.218) a.(=2.l288) s ——> NP VP (Aux) (wa) b. VP -—-> Quant c. (S) (Quant) (Det) N) N? —’Is (NP) Rule (2.218b) may be conjoined with rule (2.128b) and (2.1280) into: (2.219) (NP) V (Neg) VP —) Neg Quant A striking parallel is observed between the relevant rules that derive two types of negation and those that derive two types of quantifiers, namely, between (2.128b) and (2.218b), and (2.1280) and (2.2180). Turning to the underlying structure (2.217), this structure will go through the following stages to derive sentence (2.213): i. Application of Sentence-raising, giving (2.220) from (2.217): NP\\\\\‘ \:\\\\\ Aux VP T NP 1 Neg past Quant sono kuragfi gakusei sonkei- nai minnano the class student sono suru not all reSpect no g8 that in sensei 0 teacher ii. Application of Quantifier-attachment, giving (2.2218) or (2.22lb) from (2.220): (2.221) a. / 81P\m /\ Neg past sono kurasu Quant sonkei- nai no suru sono minL ano gakusei sensei o b. 'f””””’,—S\ :-\_‘§I__;;~§‘u‘§~§“-~Aux \ I sono kurasu Quant sonkei— n11 no I suru gakusei minna sono sensei ga 0 Turning next to the transformations used in the above derivation, the first one, Sentence-raising, is the same rule 104 that is used in the derivation of sentences containing sentential negation. To cover the derivation Of (2.220) from (2.217), the Sentence-raising rule in (2.180) needs only a slight revision such as: (2.222) 0151‘ [MEX] N]. -—>[M x N] S2 b1 1 where M, N and A are variables, and M or N S dominates VS'T or exhaustively dominates Adjs, Adv Adv N' F’ AdVR-P’ angg or Neg A new transformation is Quantifier-attachment, which may be compared with Negative-attachment. As a first approximation, to be revised later on, we may give Quantifier-attachment as: (2.223) X N Y V Z Quant‘-+ X Quant+N Y V Z where X, Y and Z are variables, and X contains no Quant or Neg We note that in Japanese most quantifiers can not only precede but also follow the nouns which they modify. Thus, the application of Quantifier-attachment to (2.220) gives (2.2218) first, and the Optional shift of the quantifier derives (2.221b). The structure (2.221b) generates sentence (2.213), while (2.2218) gives an ungrammatical sentence, such as: 105 (2.224) *sono kurasu no minnano gakusei g8 sono sensei the class in all student that teacher 0 sonkeisi nakatta. respect not-did 'All the students in the class did not reSpect that teacher.’ This is due to the idiosyncrasy of minna "all." If the quantifier is subete "all" in place of gIgga, it can both precede and follow the nouns which it modifies. Thus, both (2.2258) and (2.225b) are grammatical sentences: (2.225) a. sono kurasu no subeteno gakusei ga sono the class in all student that sensei o sonkeisi nakatta. teacher respect not-did 'All the students in the class did not respect that teacher.‘ b. sono kurasu no gakusei g8 subete sono the class in student all that sensei o sonkeisi nakatta. teacher reSpect not-did 'All the students in the class did not respect that teacher.‘ Incidentally, it should be mentioned that a quantifier is not just shifted to a proper place but is attached to a noun in that an attached quantifier and a noun together form a unit of a noun phrase just as an attached negative and a verb make a unit of a verb phrase. This is confirmed by the passive and the reflexive tests which are, according to Jacobs and Rosenbaum (1968), relevant to the discovery of noun phrases, such as: 106 (2.226) subeteno gakusei all student sono sensei wa sono kurasu no gakusei subete student all that teacher the class in ni sonkeisare nakatta. by respected not-was 'By all the students in the class, that teacher was not respected. (= That teacher was not respected by any student in the class.)‘ (2.227) subeteno gakusei g8 all student sono kurasu no zibunzisin gakusei g8 subete the class in themselves student 811 o kitaete iru. training are 'All the students in the class are training themselves.‘ Continuing the discussion of Quantifier-attachment, let us consider and compare the following pair of sentences: (2.228) ookuno hitobito ga daigakusei O sinyoosi nakatta. many people college trust not-did student 'Many people did not trust college students.‘ (2.229) hitobito wa ookuno daigakusei O sinyoosi nakatta. peOple many college trust not-did student 'Many college students, people did not trust.‘ According to this analysis, sentences (2.228) and (2.229) will be derived from (2.230) and (2.231), respectively: 107 (2o230)1M/S\ £-lyy~“\§~‘-“~“~A Quant W,,r”””fl VP I ookuno f m./”///'\‘\\\‘\\ AI many Ne past I \ hitobito ga N sinyoosuru nai people I trust not daigakusei 0 college student Sl NP’f/l/IIZ’ ~~~7777\7775‘“VT E2 QuLnt NPdr/r/If/I’ ::::::;P~“M~H~Mhfl~fifxux ooLuno I NP”/////’I \\\\\\Neg Fist hitobito g8 N sinyoosuru nai (2.231)(=2.23o) daigakusei 0 We note, first of all, that (2.230) and (2.231) are identical. This is clearly a problem since these structures cannot Offer an explanation for the semantic difference between sentences (2.228) and (2.229). Now, considering the derivation involving (2.230), after the raising Of S into 2 31' Quantifier—attachment correctly assigns ookuno "many" to hitobito "people," giving sentence (2.228). But the trouble is that this Quantifier-attachment rule cannot assign ookuno 108 "many" to daigakusei "college student" in (2.231) to generate sentence (2.229), since the leftmost constituent combinable with ookuno "many" is hitobito "people," not daigakusei "college student." Thus, there is no chance of sentence (2.229) being generated from (2.231). This poses another problem here. Moreover, if Passivization optionally applies in the 82-cycle of (2.230), the following structure is derived: (2.232) [[[daigakusei ga hitobito ni sinyoosare college student people by trusted nakattaJS JNP [[OOkunOJQuantJVPJSl 2 not-were many Then, Quantifier-attachment attaches ookuno "many" to daigakusei "college student," yielding sentence (2.233), not (2.234), though (2.234), not (2.233), is synonymous with (2.228): (2.233) ookuno daigakusei ga hitobito ni sinyoosare many college student people by trusted nakatta. not—were 'Many college students were not trusted by people.‘ (2.234) daigakusei wa ookuno hitobito ni sinyoosare college student many peOple by trusted nakatta. not-were 'By many people, college students were not trusted.‘ This presents still another problem for the Quantifier— attachment rule in (2.223). 109 Our suggestion for these problems is to make use of a dummy symbol A. More specifically, the structures (2.230) and (2.231) underlying (2.228) and (2.229) will be revised to (2.235) and (2.236), respectively: (2.235) /sl\ NP VP I l 82 Quant \V\ I NP P Aux ookuno Quant N NP V Neg past A hitobito ga N sinyoosuru nai peOple I trust not daigakusei 0 college student (2.236) 31 NP’/////” ‘~“-§--~§“~‘~““VP ,¢/////’/’;2 QuLnt NP t::::::::::;;“‘-~N‘-“‘Aux ooLuno 1 NP/ \\ I N /// \\\\ \ Neg past hitobito ga Quant N sinyoosuru nai people l I trust not A daigakusei 0 college student These structures both indicate that the quantifier ookuno is outside the scope of negation. What is more, the relevant difference between (2.235) and (2.236) can provide an 110 explanation for the semantic difference between sentences (2.228) and (2.229). Thus, the use of A solves the first problem noted above regarding (2.230) and (2.231). Furthermore, the use of A is not unique here, but is used not infrequently in the theory of transformational grammar. Chomsky (1970a, 203), for instance, suggests that the underlying structure for passives is Ngeeex-E-NE—ey A: where 21,9 is an agent phrase and the first step of the passive Operation is to replace A by the subject noun phrase. Chomsky (1970a, 204) further argues that agent- postposing is simply a generalization of one of the components of the passive transformation: for instance, if agent-postposing applies to the enemy-destructioneef £22 SEEK-2X 9, it derives the—destruction e: gee Elilfhl gee eeegy. In addition, Chomsky (1970a:l97-l98, 209) prOposes that sentences such as the question ie whether John should leege and the prospects are for peace derive from the structure [ get N Qemp 1E3 Qe [,9 JEEEQ by replacement of the unspecified predicate A by the complement of the subject noun, and that this underlying structure involving A is motivated as the matrix structure for sentences such as what John egg was hurt himself, which might be derived from [it the: John hurt John]NP 22 [ e 3 Ad0pting this Pred' analysis, Akmajian (1970, 30), in discussing pseudo-cleft sentences, posits Extraction Rule such as: [S[X-A-Y]Sbe[A]]—-) [ SE X - [+PRO,+WH] — Y 38 be E A J J 111 Moreover, Bresnan (1970, 300) notes that "Chomsky (1967, 33) suggests that the underlying structure for John broke his egg, where eie indicates inalienable possession, is £222 broke A's arm. A COpying transformation would fill in A with {egg and ultimately ee." Thus, the dummy symbol A generally occurs in underlying structure and is later filled in by the application of a relevant transformation. Now, the use of A necessitates the revision of Quantifier-attachment to the operation of replacing A by a quantifier, or put differently, attaching a quantifier to A. Thus, the Quantifier—attachment rule in (2.223) will have to be revised to: (2.237) x [ A (Det) N JNP Y Quant-——+ X [ Quant (Det) Y N JNP where X and Y are variables, and X contains no Quant or Neg Notice that Quantifier-attachment is not a transformation that inserts morphological material into a lower sentence, since Sentence-raising, applied beforehand, raises a sentence containing the unspecified quantifier A into a higher sentence containing the quantifier to be attached. Turning to the derivation of (2.228) and (2.229) from (2.235) and (2.236), after 82 is raised into 81’ the Quantifier—attachment rule in (2.237) attaches the quantifier ookuno "many" to A in (2.235) and (2.236), deriving (2.228) and (2.229), respectively. Thus, the revised Quantifier- attachment rule along with the use of A solves the second 112 problem noted above regarding the derivation of (2.229). In addition, even if Passivization applies to (2.235), the resulting structure is as follows: (2.238) [[[daigakusei ga A hitobito ni sinyoosare college student peOple by trusted r nakatta]S JNP LfookunolQuantIVPJSl 2 not—were many Then, the quantifier ookuno "many" is attached to A, deriving sentence (2.234). In this way, the use of A together with Quantifier-attachment in (2.237) solves all the above-noted problems in a natural way. The use of A and the revision of Quantifier-attachment also necessitate the correSponding revision of the underlying structure (2.217) to (2.239): (2.239) /Sl\ NP VP 1 S Quant P ’/////Y Aux minLano ‘¢::::::;;7fi \\\\‘ /,‘\\\\T I all Q nt N NP V sono ua Jeg past the I I A gakusei sonkei- nai kurasu no student sono suru not class in that respect ga sensei 0 teacher Applying Sentence-raising to (2.239), we derive (2.240) rather than (2.220): 113 /‘:\ t I" (2.240) NP ‘ Neg past Quant A gaklseiA sono sonkei- nai minnano sensei suru 0 Then, the application of Quantifier—attachment derives (2.221a) first and the optional shift of the quantifier yields (2.221b), as desired. Our next step is to show that Quantifier-attachment is not an ee hee transformation that may be applied only in the derivation of sentences containing a quantifier and negation. To illustrate with a concrete example, let us observe the following: (2.241) [Mary dake ga yakusoku o mamora nakatta]S only promise keep not-did 'Only Mary did not keep her promise.‘ (2.242) [[Mary dake ga yakusoku o mamotta wake]S de only promise kept that is we nails not 'It is not so that only Mary kept her promise.‘ We note first that sentence (2.241) is not synonymous with (2.242). We further note that in (2.242) the negative eel includes in its sc0pe the lower sentence containing the nominal adverbial eeke "only.” The non-synonymity of (2.241) with (2.242) indicates that dake in (2.241) is 114 outside the scope of negation, and that as the scope of negation is the whole sentence in which it occurs, eeke "only" has originated from a higher sentence in the underlying structure. This is clearly confirmed by the fact that sentence (2.241) may be synonymous with (2.243), in which eeke is in a higher sentence than the negative: (2.243) [[yakusoku o mamora nakatta no]S wa Mary promise keep not—did that dake da]S only is 'Lit. It is only Mary that did not keep her promise.’ If this line of argumentation is acceptable, it follows that sentence (2.241) has the underlying structure of the following form: (2.244) /Sl\ NP VP /82'\ AdVN NP \VP Aux dalke / \ _. / I \ I only Adv NP V Neg past I N l I I A Mary ga N mamoru nai keep not yakusoku o promise This structure indicates that the nominal adverbial dake is outside the scope of negation. After raising 82 into 31’ we need a transformation to attach dake to A, giving sentence 115 (2.241). But we note here that this Operation is essentially the same as that of Quantifier-attachment in (2.237). Thus, to take care of the attaching of eeke "only" to A we need no new rule but a slight revision of Quantifier-attachment such as: (2.245) Quant x [ A (Det) N 3N? Y {Adv I -——> N Quant x [{Ava I (Bet) N 3N1, Y where X and Y are variables, and X contains no Quant, Ava or Neg Now, this rule can be used to attach the nominal adverbial eeke to A, deriving sentence (2.241).2S Furthermore, it may be mentioned here that nominal adverbials are attached to the unspecified nominal adverbial A in that an attached nominal adverbial and a noun or a noun phrase together form a unit of a noun phrase. This can be proved by the passive and the reflexive tests in a way quite similar to the case of quanti- fiers (see (2.226) and (2.227)). The foregoing discussion demonstrates that Quantifier- attachment is, at least, not ee gee in that it may be applied to nominal adverbials in addition to quantifiers. Furthermore, this transformation may be applied, unchanged, to English as well, as will be discussed in section 2.9.2, and this fact will further increase the independent motiva- tion of this transformation. Amplifying the discussion of Quantifier—attachment, we 116 magr go on to consider the following examples in which two . . . . 26 quantifiers are involved in a sentence: (22.246) ookuno seito ga zyugyoo de hutatuno subako many pupil class in two nest box 0 tukutta. made 'Many pupils made two nest boxes in the class.‘ (2.247) hutatuno subako ga zyugyoo de ookuno seito two nest box class in many pupil niyotte tukurareta. by were made 'Two nest boxes were made by many pupils in the class.’ We observe first that sentence (2.246), in its primary reading, is synonymous with (2.248), but not with (2.249), While sentence (2.247), in its primary reading, is synony- Inous with (2.249), not (2.248): (2.248) [[[zyugyoo de hutatuno subako o tukutta]S class in two nest box made seito ga]NP ooku ita]S pupil many were 'Lit. The pupils who made two nest boxes in the class were many.‘ (2.249) [[[zyugyoo de ookuno seito ga tukutta]S subako class in many pupil made nest box wa]NP hutatu da(tta)]S two are(were) 'Lit. The nest boxes that many upils made in the class are (were) two in number).' We see that in (2.248) hutatuno "two" cannot include ooku 117 'mmny" 2111 its scope, since the latter occurs in a higher sentence than the former, and the Opposite is the case with UL249) . The synonymity of (2.246) with (2.248) and that of(2.224f7)'with (2.249) lead us to set up (2.250) and UL251) as the underlying structures for (2.246) and (2.247), respectively: (2250) .a””/””’S l7§PP7777§§“‘-1~V I t s/I\ I L / \ I / \ many Quant N' zyugyoo de .NP V past class i /‘\\\\\\N I A seito ga Quant tukuru pupil make hutatuno subako 0 two nest box Q-ZSU Quant NP Adv~7~777“-‘-~7-7~“‘-~A ux hutatuno /\ QUant N zyugyoo de past I I class i /NPshowr the validity of (2.250) and (2.251), therefore, we must verify this hypothesis. In this connection, let us consid er the following: (2.2EX3) ookuno seito ga zyugyoo de subako o tukutta. many pupil class in nest box made 'Many pupils made nest boxes in the class.‘ Sentence; (2.253) is synonymous with sentence (2.254), which is in tttrn derived from (2.255): (2.254) [[[zyugyoo de subako o tukutta]S seito ga] class in nest box made pupil NP ooku itaJS many were 'Many were the pupils who made nest boxes in the class.‘ (2h255) [[[seito ga zyugyoo de subako o tukutta]S pupil class in nest box made seito ga]NP ooku itals pupil many were In.the structure (2.255) underlying (2.254) the quantifier FLO-fl "many" includes in its scope the lower sentence [seito ga zyugyoo de subako o tukutta]S (= pupils made nest boxes in the class). Then, the synonymity Of (2.253) With (2:254) suggests that ookuno "many" in (2.253) also inClUdeS in its scope the whole sentence [seito ga zyugyoo de subako o tukuttajs (= pupils made nest boxes in the class). Thus, it is Shown that hypothesis (2.252) is true of sentence 119 (2.253) . CDC) take another example, let us consider (2.256): (2&2565) [seito ga zyugyoo de hutatuno subako o tukutta]S pupil class in two nest box made 'Pupils made two nest boxes in the class.’ We OMSeirve that (2.256) is synonymous with sentence (2.257), which is in turn derived from (2.258): (Z-ZEVY) [[[seito ga zyugyoo de tukutta]S subako wa]NP pupil class in made nest box hutatu da(tta)]S two are(were) 'Lit. The nest boxes that the pupils made in the class are (were) two (in number).' (2°2550 [[[seito ga zyugyoo de subako o tukutta]S pupil class in nest box made subako gaJNP hutatu da(tta)]S nest box two are(were) In‘fiue structure (2.258) underlying (2.257) the quantifier hEEflag "two" includes in its scope the lower sentence [seito €3,2yugyoo de subako o tukutta]S (= pupils made nest boxes irtthe class), and the synonymity 0f (2°256) With (2°257) indicates that hutatuno "two" in (2.256) also irICludes in.its SCOpe the whole sentence in which it occurs, namely, [seito ga zyugyoo de subako o tukuttaJS (= pupils made nest boxes in the class). Thus, we see that hypothesis (2°252) is also applicable to sentence (2.256). This can be easily extended to other similar sentences containing quantifiers. In this way, hypothesis (2.252) can be eStablished as a linguistic fact observed in Japanese. 120 INCyw, let us return to underlying structures (2.250) and (2.251). Based on the above-established fact--the scope of a quantifier is the whole sentence in which it occurs-- these structures indicate that the quantifier in S1 is outsniel'the scope of the quantifier in 82. More specifi- caflyy :ir1(2.250) the scope of hutatuno "two" is 82 and cmkuno 'Fnany“ in S1 is outside of its s00pe, whereas in (2.251) 'the sc0pe of ookuno is 82 and hutatuno in S is 1 OUtSUie E M x N 33 2 l 1 where M, N and X are variables, and M or N dominates VSoT or exhaustively dominates AdjS, Ava, AdvF, AdVR.P’ Quant or Neg That 21s, the same rule is applicable to both Japanese and Englisfli, which provides additional support for this trans format ional rule . TWne second transformation is Quantifier—attachment, w“ n10h égives (2,272) from (2.271). We note here again that 126 Quantifier-attachment given in (2.245), repeated here in (2.274), can be applied, unchanged, to English quantifiers as well: (2.274)(=2.245) Quant x [A (Det) NJNP Y {Ava }—9 : Quant x [1,,va } (Det) N JNP Y where X and Y are variables, and X contains no Quant, Ava or Neg floreover, the same transformation is necessary to deriflle sentences such as (2.275) involving the nominal adVerbial _o__r_1_l_y: (22.275) ley that passenger did ngt pay the additional fare. E3en‘tence (2.275) is not synonymous with (2.276), in which 222;: is in the scope of negation: (22.276) a. It is ngt so that gnly that passenger paid the additional fare. b. N93 gnly that passenger paid the additional fare. The rmin-synonymity of (2.275) with (2.276) suggests that w in (2.275) is outside the scope of negation. This is additixonally confirmed by the fact that (2.275) may be Synondnnous with (2.277), in which ggly is outside the scope of neé—Z'ation: (2-277) It was gnil that passenger that did 292 pay the additional fare. 127 If gnly in (2.275) is outside the scope of ggt, it must be outside the sentence containing n93, since the scope of ggt is the whole sentence in which it occurs. Thus, the underlying structure for sentence (2.275) will be: (2.278) [[[[talAva [that]Det [passengerJN]NP [pastlAuX [[nOtJNeg [payJV [the additional fareJNPJVPJS2JNP [[onlylAdv JVPJS N 1 We see that this structure indicates that only in the higher sentence is outside the scope of n93. Moreover, it is clear that the derivation of (2.275) from (2.278) requires the attachment of gnly to A as well as the raising of 82 into 81’ Here, the point is that this attachment of ggly can be covered by the same rule stated in (2.274). Summarizing so far, we have indicated that Quantifier- attachment is not ad hgg in English in that it may be applied to nominal adverbials in addition to quantifiers. Furthermore, Quantifier-attachment given in (2.274) can be applied, unchanged, to English quantifiers and nominal adverbials as well as to Japanese ones. The applicability of the same Quantifier—attachment rule to both Japanese and English clearly increases the independent motivation of this transformation. Returning to the underlying structure (2.270), many in this structure will be derived by the rule (2.279b) as a sentential quantifier, as Opposed to a nominal quantifier derived by the rule (2.279c): 128 (2.279) a.(=2.l90a) s ——9~NP (Aux) VP b. VP —-) Quant (Quant) (Det) N (s) ‘NP"" ((NP) 8 I Rule (2.279b) may conjoin with (2.190b) and (2.190c) into: (2.280) (Neg) V (NP) VP —-) Neg Quant C. 27 fhflxe (2.2790) may be compared with (2.l90c) that derives Verb-phrase negation. Continuing this discussion, we may now proceed to COHSixier examples containing two quantifiers in a sentence such as: (2.;281) a. nggy child likes sgmg animals. b. §gmg animals, £1331 child likes. c. ‘§gm§ animals are liked by gyggy child. Sentennce (2.281a) is not synonymous with either (2.281b) or (2-2Eilc), and according to our analysis, the structure underlying (2.281a) will be: (2.282) s /l\ NP VP I \ I 82 Quant NP \§:-~‘\“--_VP | ux every Q \1 I / \ Ileant I pres Y <3 child like Quant N I I some animals 129 The structure (2.282) indicates that every in the higher sentence is outside the sc0pe of some, thus explaining why some in sentence (2.281a) cannot include every in its SCOpe. If no moving transformation applies in the SZ-cycle of (2J282), Quantifier—attachment applies, after 82 is raised into Sl’ to generate sentence (2.281a). If, on the other hand, Topicalization or Passivization applies in the 82- CYCle of (2.282), the resulting structure is: (22.283) a. [[[some animals, A child likes]S ]NP 2 [[EXEEXJQuantJVPJSl b. [[[some animals are liked by A childJS JNP 2 [[gzgzijuantJSl In both (2.283a) and (2.283b) Quantifier-attachment is blOCflsed since the unspecified quantifier A follows another quan-t ifier some. In this way, the Quantifier-attachment rule ‘with.the condition "X contains no Quant or Neg," as Statmed_j11(2.274), Can explain why there is no passive or topicalized counterpart of (2.281a), derived from the Structure (2.282) underlying (2.281a). In turn, sentences (2.281b) and (2.2810) are synony- mouf3, and they will be derived from the underlying structure (2‘284) under our analysis: 130 (20284) /Sl . ‘\\\\\‘\~VP A I“ / \. (Quant N pres V I I /\ child like Quant I I A animals every This Estructure indicates that some includes every in its SCOpea, but not the other way around. Now, if no moving trarmiformation applies in the 82—cycle, Quantifier- atta£fl1ment cannot apply in the Sl-cycle, since the unSEKBCified quantifier A follows another quantifier every. OrllY‘whencertain moving transformations apply in such a way as tC> prepose the unSpecified quantifier A over every can quan'tizfier—attachment apply. Thus, if Topicalization or PasSi‘vization applies in the SZ-Cycle of (2.284): the reStilising structure will be (2.2853) or (2-285b), respe ctively: (22.2385) a. [[[A animals, gyggy child likesls JNP 2 [[SomejQuantJVPJSl b. [[[A animals are liked by gygyy childls ]NP 2 [EgggnguantJVPJSl Then ’ Ei-Jpplying Sentence-raising and Quantifier-attachment, Wed - Eil‘JL‘Ue sentences (2.281b) and (2.2810), respectively. In 131 this case again, Quantifier-attachment can account for there being no non-topicalized or non—passive counterpart of (2.281b) and (2.28lc), derived from the underlying structure (2.284). 2.10. The Discussion and Solution 9f Remaining Problems Based on the analysis proposed in the preceding sectixins, we may now proceed to reexamine and solve those Intfifilems mentioned in section 2.8. Let us consider each of them: A. Ikccording to the proposed analysis, sentences (2.200), (2.201) and (2.212), repeated here as (2.286), (2.287) and (2.288), will have underlying structures (2.289), (2.290) alki (2.291), respectively: (2.286)(=2.2oo) M gggy arrows hit the target. (2.287)(=2.2Ol) giggy arrows did M hit the target. (22.288)(=2.212) The target was not hit by many arrows. NP’////””’ l‘\\\\\\\\\‘ ../ . (2.289) VP I Neg I NP’/I’II’AE:‘\\\\\\‘\VP not Quant\\\‘N pist V////\\\SNP I I \ many arrows hlt D T the target 132 NP’//////’Sl\\\\\\\\‘VP AZ QJant ///ZVP\ (2.290) Aux many /\ /IPNP\ Quant past Neg I 1 , /\ A arrows . n t hlt Det N the target (2.29l)(=2.289) ’/////,sl\\\\\\\\\ NP VP I I ”’,,/””SZ Nfg NP Aux/ ‘\\\\VP no \ I /\ Quant N past V NP many arrows hit Df4‘\\\N the talget Wee riots first that the structure (2.289) underlying (23-2286) and the structure (2.291) underlying (2.288) are jxietntical. This provides an explanation for the synonymity OI? ( 2.286) and (2.288). Considering (2.289) first, if no ”“3‘71J1g transformation applies in the S -cycle, Sentence- 2 I“Bdlsing and Negative-attachment apply, yielding sentence (22°23536). If, on the other hand, Passivization applies to (23 "2353EH, the resulting structure is as follows: 133 (2.292) [[[the target was hit by many arrows]S JNP 2 [EQQEJNngVPJSl Applying Sentence-raising and Negative-attachment to attach got to the verb, since the verb follows no quanti- fier or negative in this case, we get sentence (2.288). Thus, this approach can account for the fact that Sentence (2.288) is synonymous with (2.286), since they are derived from the same underlying structure, (2.289): in particular, the structure (2.289) involving go_t_ in a higher sentence than may can explain why _me_1_ny in (2.286) and (2.288) cannot include £1.91; in its scope. Furthermore, this approach keeps intact the "meaning-preserving" definition of transformations such as Passivization and Topicalization. Considering next (2.290), if no moving transforma- tion applies in the Sz-cycle, Sentence-raising and Quantifier-attachment apply, giving sentence (2.287). If, On the other hand, some moving transformation, for irlStance, Passivization applies in the S2-cycle of (2.290), the following structure is derived: (2.293) [[[the target was not hit by A arrows]S 3NP 2 [CMJQuantJVPJSl Then, Quantifier-attachment is blocked in the Sl-cycle, Since the unSpecified quantifier A follows the negative m. In this way, our approach can explain why there is no Passive counterpart of (2.287) that is derived from 134 (2.290). Furthermore, we noted in section 2.8 that if the structure underlying (2.287) is (2.203) (repeated here as (2.294)) instead of (2.29o), then there is no reason to block the application of Passivization to (2.294), giving senatence (2.288), though it is not synonymous with (2.287): (2.294)(=2.203) [[2221 arrOWSJNp [paStJAux [EEQEJNeg [hit]V [the targetJNPJVPJS Tllis problem is solved if we adOpt (2.290), not (2.294), as; the underlying structure for sentence (2.287). That is tc> say, Passivization may freely apply in the 82—cycle of (22.290), since there is no reason to block its applica- tixon, but if Passivization applies there, Quantifier- Erttachment is blocked in the Sl-cycle of (2.290), since tile condition of this transformation is not satisfied in tkuat case. Thus, the filtering function of Quantifier— airtachment correctly blocks the derivation of sentence (23.288) from (2.290) underlying (2.287). This gives additional support to the underlying structure (2.290). Next, let us consider sentences (2.204), (2.205) and (2.:206), repeated here as (2.295), (2.296) and (2.297), respectively: (22.295)(=2.204) The police did n9; arrest ggpy demonstrators. (2. 296)(=2.205) Many demonstrators were not arrested by the police. 135 (2.297)(=2.206) Not many demonstrators were arrested by the police. Under our analysis, their underlying structures will be (2.298), (2.299) and (2.300), respectively: NP”////”’Sl‘\\\\\\\“VP I I l/I:\ I: ’//§E\\ I V’///' \\\\\NP Det N past (2.298) tLe police arrest Quant N I many demonstrators (2.299) s "/”/,,,. 1‘\\\\\\\\‘ TP WP S Quant / l 2\ l NP Aux VP many ///M\\\ I ’/’//<;? \\\\\\ Det N past Neg Y NP the police nLt arrest ngg; \\\\\N I A demonstrators 136 (2.300)(=2.298) fil‘~\\\\\\\\ NP VP I /82 mfg NP Au; ‘\\\\\\\‘VP not .///l\\\ I ’///’ ‘\\\\ Det N past V NP i I. I H //// \\\\\ t e po lce arrest Quant N many demonstrators We: note first that (2.298) underlying (2.295) is identical quth (2.500) underlying (2.297), which provides an expla— Iuation for the synonymity of (2.295) with (2.297). We nxrte next that the structure (2.298) underlying (2.295) alui (2.297) indicates that nay in the higher sentence is Otrtside the s00pe of many, thus explaining why in (2.295) azui (2.297) many cannot include nay in its s00pe. Now, lxyt us assume tentatively that the underlying structure ifOI'(2.295) is not (2.298), but (2.301) involving verb- Iflrrase negation in place of sentential negation in (2.298): (2.501) [[the police]NP [pastJAuX [Engtheg [arrest].V [many demonstratorslNPJVPJS T1len, we cannot account for the synonymity of (2.295) with (23-2297) in a natural way. Moreover, we cannot explain why irl ESentence (2.295) many cannot include nay in its scope. The Scope of many is the whole sentence in which it (DCC311113, so if (2.301) underlies sentence (2.295), many 8 11C3ttl the verb, since the verb follows no quantifier or IUBgative in this case, giving sentence (2.295). 0n the Otflqer hand, if Passivization applies to (2.298), the Iwesulting structure is: (2.302) [[(many demonstrators were arrested by the Alxplying Sentence-raising and Negative-attachment, we (iezdve sentence (2.297). Thus, this approach can account Zfox‘the fact that sentence (2.297) is the passive Cx>unterpart of (2.295), derived from the same underlying Structure , (2. 298) . Turning next to (2.299), if no moving transformation aPuIlies in the Sz-cycle, Quantifier-attachment is blocked: thf? unspecified quantifier A follows the negative nay, ‘Whith does not meet the condition of Quantifier- attachment. If, on the other hand, Passivization applies ill ‘the S -cycle of (2.299), the following structure is 2 deI‘ived: (23.303) [[[A demonstrators were not arrested by the policeJSZJNP [EmanijuantJVPJSl 138 After 82 is raised into 81, Quantifier—attachment can In apply to attach many to A, giving sentence (2.296). this case, too, this approach can explain, in terms of the filtering function of Quantifier—attachment, why there is no non-passive counterpart of (2.296), derived from the Inaderlying structure (2.299). In section 2.8, we further noticed that (2.208), Iwepeated here as (2.304), is not acceptable as an tuaderlying structure for (2.205), repeated here as (2.305): (2.504)(=2.208) [Ithe policeJNp [pastJAux [EEQEJNeg [arrest]V [many demonstratorslNPJVPJS (2.305)(=2.205) Many demonstrators were not arrested by the police. (2.306)(=2.204) The police did not arrest many demonstrators. Vte noted there that when Passivization does not apply to (2.304) underlying (2.305), the resulting sentence is (22.506), although (2.506) is not synonymous with (2.505); OILly when Passivization applies to (2.304) will sentence (23.305) be derived. Thus, there is no natural way to blxbck the derivation of (2.306) from (2.304), if the ‘uIKierlying structure for (2.305) is (2.304). This FNPCYblem simply does not occur, if (2.299), rather than (23-2304), is adopted as the underlying structure for SEEIFtence (2.305), as proposed here. Consider once again (’23-2299), repeated here as (2.307) for convenience of 139 reference: (2.507)(=2.299) s ’//////”’ l“\\\\\\\\\‘VP TP Quant /:32 \ NP Alle VP maIny \\\ I ,z”//<7 ~\\\\\~I N past Neg V NP I / \N not arrest Quant A demonstrators Ikpplying Sentence—raising, we derive (2.308): (2.308) NP"””ZNEE:::;7, l\\V1> P \\\‘ I ”///:;;/I\\\\\\‘NP QIant Det N past Neg ‘ l arrest Quan:/‘\\\\N mJny the police A demonstrators III order to derive (2.306) from (2.308), many must be But this airtached to A preceded by the negative not. OEHBIetion clearly violates the condition of Quantifier- In other words, such an operation is blocked at 1: achment . tllrwaugh the filtering function of this transformation. Underour analysis, therefore, there is no chance of Sfierrtence (2.306) being derived from (2.307) underlying (2-3om. 140 B. The first type of counter-example such as (2.210), as already discussed at the end of section 2.8, poses no problem in our analysis. But the second type of counter-example such as (1.12), repeated here as (2.309), iJTvolving extra heavy stress on the quantifier is still a Ixroblem within our framework: (2.309)(=1.12) The students did 223 read many books. Wt; may only suggest one possible solution for this problem, tuased on an assumption proposed by Bresnan (1971a), Ekierwisch (1968) and Pope (1971). Bresnan argues that Cxertain phonological rules such as the Nuclear Stress Rule ease ordered after all the syntactic transformations on Enach transformational cycle. Bierwisch has reached a (nonclusion that in certain German sentences, stress must IDe assigned or somehow marked before some of the syntactic lulles apply. Similarly, Pope argues that some phonological Ifllenomena such as intonation assignment must precede some SYTmactic deletion rules. What is common to their argu- Iments is that syntax and phonology interpenetrate. If we 'aSENLme, following them, that some phonological rule that affiiigns extra heavy stress on quantifiers applies before Qtlantifier-attachment applies, a slight revision of Qluirndider-attachment may take care of examples such as (£3.2309). This revision will be something like: (23.310) Quantifier-attachment may attach a quantifier to A preceded by another quantifier or a negative, if the quantifier to be attached has 141 extra heavy stress. lhis proviso may allow Quantifier-attachment to apply to structures such as (2.312), which is in turn derived from the underlying structure (2.311) by the application of Sermence-raising and the stress-assignment rule: (2.311)Il‘/~1\VIP Quant NP/ A\:x\\ VP maLy ’///"\\ I "’///:%/ \\\\‘NP I I not read Quant N I A books Dé€///N§\\N' :fsthg /// /\\\\\W Q3:nt I Neg! I I\ ,I the students not read Quant T many A books ‘AEflDlying Quantifier—attachment to (2.312), we attach ___y 't0 4Q” though it follows nay, deriving sentence (2.309) C’ This analysis, as already mentioned in section 2.8 does nfi2t involve the application of the Quantifier-lowerlng ITULLE? which has been criticized as an 22 nan transformation 1 r1 'tllat it violates a presumably universal condition on 142 transformations. Instead, this analysis makes use of Quantifier—attachment, Sentence-raising and Negative- attachment, which are all shown to be independently moti- vated transformations in both Japanese and English. D. The underlying structures proposed by this analysis are free from those criticisms which are made against underlying structures such as (1.3) and (1.4) proposed by G. Lakoff. The underlying structures within our framework are well—motivated in that: 1. The underlying structures and their corresponding surface structures are related by means of inde- pendently motivated transformations in both Japanese and English. The underlying structures do not involve the relative clauses that have been criticized as unique by Chomsky (1972, 183). The underlying structures are based on putative linguistic universals regarding the scope of negation and that of quantifiers. Taking the argument of this section one step further, we may proceed to a discussion of the examples in which apparently only one quantifier is involved. To take concrete examples, consider the following: (2.313) a. Some pupils made a nest box in the class. b. A nest box was made by some pupils in the class. (2.314) a. Many students know a foreign language. 143 b. A foreign language is known by many students. There is a clear semantic difference observed between the (a) and the (b) sentences in (2.313) and (2.314): (2.313a) means that there are some pupils who (each) made a nest box in the class, while (2.3l3b) means that there is one nest box that some pupils made in the class; similarly, only (2.3l4a) has the reading that there are many students who know a foreign language, while (2.314b) means that there is a (particular) foreign language that many students know. In other words, the s00pe of mama is the whole sentence in (2.313a), but not in (2.3l3b), and the scope of many is the whole sentence in (2.3l4a), but not in (2.3l4b). In comparison with (2.313) and (2.314), consider next the following examples containing the definite determiner gnag in place of a in (2.313) and (2.314): (2.315) a. §nma pupils made inay nest box in the class. b. inan nest box was made by anma pupils in the class. (2.316) a. Many students know gag: foreign language. b. Tna: foreign language is known by many students. It should be observed that the (a) sentence is synonymous with the (b) sentence in (2.315) and (2.316): both (2.315a) and (2.315b) mean that there are some pupils who made that nest box in the class, and (2.316a) and (2.3l6b) have the reading that there are many students who know that foreign language. Accordingly, the s00pe of some is the whole 144 sentence in both (2.315a) and (2.315b), and so is the s00pe of many in (2.3l6a) and (2.316b). The above comparison of (2.313) and (2.314) with (2.315) and (2.316) suggests that the semantic difference between the paired sentences in (2.313) and (2.314) is due to the indefinite article a in them. If we treat a as a kind of numeral meaning ”one,"28 namely, a kind of quanti- fier, then the above—observed semantic difference can be explained in our framework: the structures underlying (2.313a), (2.313b), (2.3l4a) and (2.3l4b) will be (2.317a), (2.317b), (2.318a) and (2.3l8b), respectively: (2.317) a. /81\ NP VP ’/’//,r"?2N~“‘-2~n““~“§“-‘- Qu%nt NP Aux \ VP Adv some Quant T past V /NP\\\\ A pupils make Quant N in the cla§s I a nest box NP/EMAM Ia Quagt \\\\N pgst ‘V///I\\\\NP Z///h\i;\\\\ I I I / \ N in the class some pupils make Quant A nest box (20318) a' /S]_\ NP VP I NP’//////’3:;\\\\\\\\‘VP ::::t /\N I V/ \N? Quint ‘ pres I A students know Quah;//’ I a foreign language NP VP \ a Quant N pres I maly stugents know QualEr/lt:;;/\\\\\\\\\\\\> A foreign language 146 The structure (2.3l7a) indicates that mama in the higher sentence includes a in its s00pe, but not the other way around, thus explaining why mama in (2.313a) can include a in its s00pe. 0n the other hand, (2.317b) shows that a in the higher sentence is outside the 800pe of mama, thus explaining why mama in (2.313b) cannot include a in its scope. Thus, the relevant difference between (2.3l7a) and (2.317b) provides an explanation for the semantic difference between (2.313a) and (2.3l3b). Similarly, the difference between the structures (2.318a) and (2.3l8b) reflects the semantic difference between (2.3l4a) and (2.3l4b). Furthermore, intuitively, the semantic difference between (2.5l5a) and (2.313b) or (2.3l4a) and (2.3l4b) seems to be quite parallel to the semantic difference between the following paired sentences involving two quantifiers: (2.319) a. Eyamyone in the room knows yya languages. b. Tya languages are known by ayanyone in the room.29 (2.320) a. Mani children like mama animals. b. mama animals are liked by many children. Under our analysis, they are derived from the underlying structures (2.321a), (2.32lb), (2.322a) and (2.322b), reSpectively: 147 NP/01\VP I I S Quant NP//’ 2.n::::::1:§77‘7\77"VP eJery Qu:nt4: pres V/ \ NP I / \ on in the room know Quant N (2.321) a. two languages I I S Quant Quag \Mles V/ \NP I I I I\ every one in the room know Quant N I A languages (2.322) a. /81\ I NP VP S Quant NP Aux VP many ,./\ \ /\ Quant N pres 1 I I chiIdren like Quant N I I some animals 148 l~\\\\\\\\‘ NP VP ’////”/,;2 Q%ant l ~\\\\\‘\~ NP Aux VP some / \ I / \ Quant N\ pres V NP many children like Quant N A aniials Note the striking parallel of (2.317) and (2.318) to (2.321) and (2.322) with respect to quantifiers and their sentential hierarchy. If we adopt the underlying structures (2.317) and (2.318), we may treat the semantic difference between the paired sentences in (2.313) and (2.314) in a way quite similar to that between the paired sentences in (2.319) and (2.320). Moreover, the derivation of (2.313) and (2.314) from (2.317) and (2.318), respectively, can be covered in terms of Sentence-raising and Quantifier-attachment in a way similar to the derivation of (2.319) and (2.320) from their underlying structures. To take up (2.3l7a) first, if no moving transformation applies in the S2-cycle, Sentence- raising and Quantifier-attachment apply, giving sentence (2.313a). If, on the other hand, Passivization applies to (2.317a), the resulting structure is as follows: (2.323) [[[a nest box was made by a pupils in the ClaSSJS JNP [[SomBJQuantJVPJS 2 l 149 Then, Quantifier-attachment is blocked since the unspecified quantifier A follows another quantifier (a in this case). Thus, the same Quantifier-attachment rule can correctly predict that there is no passive counterpart of (2.313a) derived from the underlying structure (2.317a), in addition to covering the derivation of (2.313a) from (2.317a). Turning next to (2.317b), if no moving transformation applies, Quantifier—attachment is blocked because the unspecified quantifier A to which a is to be attached follows another quantifier sgmg. But if Passivization applies to (2.317b), the following structure is derived: (2.324) [[FA nest box was made by EQEE pupils in the class]s2]NP [FQJQuantJVPJSl Then, applying Sentence-raising and Quantifier-attachment, we derive sentence (2.3l3b). Thus, Quantifier-attachment can take care of the derivation of (2.3l3b) from (2.317b). Moreover, it can correctly block the derivation of the non- passive counterpart of (2.313b) from (2.317b). A similar argument holds for the derivation involving (2.318a) and (2.318b). Applying Sentence-raising and Quantifier-attachment to (2.318a), we generate sentence (2.314s). But if Passivization applies to (2.318a), Quantifier—attachment is blocked, giving no grammatical sentence. In the case of (2.318b), if Passivization applies in the Sz-cycle, the application of Sentence-raising and Quantifier—attachment yields sentence (2.3l4b). But if no moving transformation applies in the 82-cycle of (2.318b), 150 then Quantifier-attachment is blocked, giving no grammatical sentence. In this way, the same Quantifier-attachment rule can correctly predict that there is no passive counterpart of (2.3l4a) derived from (2.318a), nor any non-passive counterpart of (2.3l4b) derived from (2.318b). Next, we wish to consider examples of a different type such as: (2.325) jany arrows did ngt hit targets. (2.326) Many targets were not hit by arrows. The underlying structures for these sentences are (2.327) and (2.328), respectively: NP/Sl\VP I NP/ 2\uX\‘VP / \ \ / Quant N past Neg V NP I I I I A arrows not hit N targets (2.327) Quant many 151 (2.328) /sl\ NP VP I I 82 Quant NP Aux VP many I ,////77 ‘\\\\ N past Neg V NP 1 I / \ arrows not hit Quant N A targets The relevant difference between these structures reflects the semantic difference between sentences (2.325) and (2.326). Now, considering (2.327), if no moving transfor- mation applies in the S2—cycle, Sentence-raising and Quantifier-attaohment apply, giving sentence (2.325). On the other hand, if Passivization applies to (2.327), the resulting structure is as follows: (2.329) [[[targets were 223 hit by A arrOWSJS JNP 2 FFEQEXJQuantJVPJSl Then, Quantifier-attachment is blocked since the unspecified quantifier A follows the negative nay. Furthermore, the use of A correctly blocks the attachment of nany to targets in (2.329) to derive sentence (2.326). In the case of (2.328), if no moving transformation applies in the 82-cycle, Quantifier-attachment is blocked because A follows the negative nan. In particular, the use of A can block the attachment of many to arrows to give sentence (2.325). If, on the other hand, Passivization applies to (2.328), the 152 following intermediate structure is derived: (2.330) IIEA targets were not hit by arrows]S 3NP 2 [EEQQXJQuantJVPJSl Then, the application of Sentence-raising and Quantifier- attachment yields sentence (2.326). In this way, the use of A is necessary not only to distinguish the structure underlying (2.325) from that underlying (2.326) but also to take care of the derivation of sentences (2.325) and (2.326) from (2.327) and (2.328), respectively, correctly blocking the derivation of (2.326) from (2.327) and that of (2.325) from (2.328). To make this point clearer, let us compare (2.327) and (2.328) with (2.331) and (2.332) which are (2.327) and (2.328) with A missing: (2.331) ([Iarrows past not hit targets]S JNP 2 [[QQEEJQuantJVPJSl (2.332) rIIarrows past not hit targets] 1: ——— 82 NP [EmanijuantJVPJSl We note first that (2.331) and (2.332) are identical. This indicates that without the use of A, no distinction can be made between the structure underlying (2.325) and that underlying (2.326), though (2.325) and (2.326) are not synonymous. Furthermore, assuming that (2.331) is the underlying structure of (2.325), the application of Passivi- zation and Sentence-raising gives the following intermediate structure: 153 (2.333) [targets were not hit by arrows [Iggflxjouantjvrjsl Then, there is no natural way to block the attachment of many to targets which derives sentence (2.326). Thus, this analysis allows the derivation of sentence (2.326) from the structure underlying sentence (2.325), though (2.326) is not synonymous with (2.325). Furthermore, if (2.332) is the underlying structure of sentence (2.326), there is no reason to block the attachment of many to arrows which gives sentence (2.325). In this case again, this analysis permits the derivation of sentence (2.325) from the structure underlying sentence (2.326), in spite of the fact that sentence (2.325) is not synonymous with (2.326). Notice that all these problems are solved or simply do not occur if we make use of A, as seen above. The foregoing argument regarding sentences (2.325) and (2.326) clearly demonstrates that the use of A is necessary to take care of the correct derivation of these sentences. This argument remains true of sentences involving nominal adverbials, which will be treated in the next chapter. Lastly, let us consider sentences such as (2.334) and (2.335) Cited from Langacker (1972, 234): (2.334) a. n9: many of the boys didn;m_consult John. b. *John wasnlm consulted by nay many of the boys. (2.335) a. n9; many of the students werenLy examined by the doctor. 154 b. *The doctor didan examine nay many of the students. If his judgement about the grammaticality of these sentences is acceptable, grammatical sentences (2.334a) and (2.335a) are derived, under our analysis, from (2.336) and (2.337), respectively: (2.336) SZ/Sl\ NP VP /3A\~ NI;~\\\ii\\‘\~A p::t %é////)K:\ many Quant D t N l oys not congultN N th b (D John (2.337) Neg quf\‘VP I not I I S Quant NP Skux A /yP many th\ I past Nfg Y the doctor not examine Quant et A the students Considering (2.336) first, after S3 is raised into 82 and 155 ‘1 the quantifier many is attached to A, S2 is further raised into Sl.‘ Then, the negative is attached to the quantifier, yielding sentence (2.334a). If, on the other hand, Passivization applies in the S -cycle, the following 3 intermediate structure is derived: (2.338) ’///,,,sl 4 NI 8% ,/’/’ 2 NP \VP n|ot S QJant NP VP many I John was not consulted by A of the boys Then, after S3 is raised into 82, the application of Quantifier-attachment is blocked since A follows the negative nay. If we attach many to A, violating the condition of Quantifier—attachment, we will get an ungram- matical sentence such as: (2.339)(=2.334b) *John wasnL: consulted by nay many of the boys. We note that the generation of (2.339) from (2.338) also violates the condition of Negative—attachment in that it involves the attachment of not in S to many which is l preceded by another negative not. —Thus, the generation of ungrammatical sentences such as (2.339) is correctly blocked, in our analysis, through the filtering function of the two 156 transformations. In other words, there is no chance of sentences such as (2.339) being generated in our analysis. Turning to (2.337), after S is raised into S the 3 2’ application of Quantifier-attachment is blocked since A is preceded by the negative nay. Thus, there is no chance of ungrammatical sentences such as (2.340) being generated in our analysis: (2.340)(=2.335b) *The doctor didniy examine nan many of the students. The generation of (2.340) from (2.337) violates the ‘condition of Negative-attachment as well as that of Quantifier—attachment, since it involves the attachment of £21 in S1 to many which is preceded by the negative nay. Thus, the filtering function of these transformations correctly blocks the generation of ungrammatical sentences Such as (2.340)- Now, turning back to (2.337), if Passivization applies in the SB-cycle, it gives the intermediate structure (2.341): NP VP 3' N‘ 2 95 Ilflj/ \VP nlot S Quant //’ 3 NP ‘\\\~\‘~‘VP mgny A of the students were not examined by the doctor Then, after 33 is raised into 32, Quantifier-attachment 157 applies to attach many to A. Next, the application of Sentence-raising and Negative-attachment to attach nay to many yields sentence (2.335a). In this way, our analysis can correctly predict the ungrammaticality of sentences such as (2.334b) and (2.335b) as well as generate grammatical sentences such as (2.334a) and (2.335a). 2.11. Conclusions In this chapter we have discussed the interrelations of negation and quantifiers. The major findings of this chapter may be summarized as follows: 1. The proposed analysis of negation and quantifiers is based on the two putative universal we noted regarding the scope of negation and that of quantifiers: a) The sc0pe of negation, sentential or verb-phrase, is the whole sentence in which it occurs. b) The scope of a quantifier, sentential or nominal, is the whole sentence in which it occurs. 2. This analysis involves the twofold distinction of nega- tion, sentential and verb-phrase, and that of quantifiers, sentential and nominal. Sentential negation and quanti- fiers are derived by the rule that rewrites VP and they "command" the sentences they modify in the underlying structures. In contrast, verb-phrase negation and nominal quantifiers are derived by the rules that rewrite VP and NP, respectively, as their optional constituents. 3. 158 Thus, this analysis contains the following base rules in Japanese and English: Japanese: S -—9’NP VP (Aux) (wa) (NP) V (Neg) VP -—) Neg Quant (s) (Quant) (Det) N} NP‘—9 s (NP) English: 8 ——+ NP (Aux) VP (Neg) V (NP) VP -—-)‘Neg Quant (Quant) (Det) N (S) (NP) S } NP——.{ Furthermore, this analysis requires some transformations to map the underlying structures derived by the above base rules into the corresponding surface structures. These are Sentence-raising, Negative-attachment and Quantifier-attachment in both Japanese and English, and Contrastive ya-attachment in Japanese, which are all shown to be independently motivated transformations. In particular, Sentence-raising and Quantifier-attachment have the same form and are applicable to both Japanese and English. The analysis in terms of these base rules and transfor- mations is applicable to both Japanese and English, and has several other advantages over both the interpretive and the generative-semantic analyses of negation and quantifiers. CHAPTER II FOOTNOTES 1. Unless otherwise stated, all the sentence-examples are the author's. Moreover, Japanese sentences are given in a transcription which shows as closely as possible one- to-one correspondence between Japanese and English, particularly in regard to the treatment of Japanese particles. 2. Another Japanese equivalent of only is dake, and so the following may also be the counterpart of (2.10): (a) sono syoonen dake ga kareno hahaoya c that boy only his mother settoku-dekiru. persuade can 'Only that boy can persuade his mother.’ But there are some cases in which sika . . . nai can be used but gang cannot, for instance, as in the following: (b) kyoo wa zikan ga zyuubun atta ga sono today time plenty of existed but that hon sika yome nakatta. book only can-read not-past 'I had plenty of time today but I could read only that book.‘ If we replace sika . . . nai with dake, the resulting 159 160 sentence will be unacceptable, or at least very awkward: (c)?*kyoo wa zikan ga zyuubun atta ga sono today time plenty of existed but that hon dake yometa. book only could-read 'Lit. I had plenty of time today but I could read only that book.’ Thus, sika . . . nai and dake together may correspond to the English only in all its environments. Notice that siru is the basic form, namely, the present indicative form of the verb and it takes the form sira when followed by the negative nai. Sentence (2.13) is derived from the structure (a) by replacing the subject marker ga by the tOpic marker ya: (a) [Ikare ga kinoo tosyokan de benkyoosita wake]S he yesterday library in studied that 2 de wa nai]S is not 1 This replacement also occurs in the derivation of simplex sentences such as (2.12); sentence (2.12) derives from the structure (b) by the replacement of ga by XE: (b) [kare ga kinoo tosyokan de benkyoosi nakatta]S he yesterday library in study not-did The replacement of ga by ma is, however, not without restriction. For instance, it cannot occur in relative clauses as in (c): 161 (c) [[[haha {pig} tukuttaJS suupuJNP wa oisiiJS mother ‘ made soup delicious 'The soup that my mother made is delicious.‘ The ungrammaticality of (c) with ya in place of ga shows that the topic marker ya cannot occur in certain kinds of embedded sentences. Based on this observation, one might argue that the surface structure of sentence (2.13) will be (d) rather than (e): (d) [kare wa [kinoo tosyokan de benkyoosita wake]S he yesterday library in studied that 2 de wa nai]S l is not (e) [[kare wa kinoo tosyokan de benkyoosita wake]S 2 he yesterday library in studied that de wa nails is not 1 Whichever may be the case, however, it is a matter of surface structure and it does not influence the fact that kama ya in sentence (2.13) occurs in 82, not in 81’ in the underlying structure. Therefore, even if we adOpt (d) rather than (e) as the surface structure of (2.13), it does not influence our discussion here. The same is true of the other sentences of the (2.13) type in this thesis. It seems that degree adverbials can co-occur with compound adjectives containing the negative nai when they are synonymous with their positive counterparts such as: 162 (a) syooziki de nai 'not honest' = husyooziki (da) 'dishonest' yoku nai 'not good' = warui 'bad' omosiroku nai 'not interesting' 2 taikutu (da) 'dull' kirei de nai 'not clean' = kitanai 'dirty' 0n the other hand, compound adjectives that cannot co-occur with degree adverbials have no such positive counterparts. For instance, consider the following: (b) husyooziki de nai 'not dishonest' # syooziki (da) 'honest' tiisaku nai 'not small' # ookii 'big' ookiku nai 'not big’ # tiisai 'small' minikuku nai 'not ugly' # utukusii 'beautiful' It is beyond the sc0pe of this thesis to set up a constraint incorporating all these observations. But the point is that the compound adjectives in (b) cannot co-occur with degree adverbials and so sentences involving them such as (2.28), (2.30), (2.32) and (2.33) become unambiguous owing to this restriction. Incidentally, observe the following: (o) i. kare wa sonnani husyooziki de nai. he so dishonest is not 'He is not so dishonest.‘ ii. kare wa ammari husyooziki de nai. he too dishonest is not 'He is not too dishonest.’ 163 iii. kare wa manna husyooziki de nai. he more dishonest is not 'He is not more dishonest.‘ Since sonnani "so," ammari "too" and manna "more" are degree adverbials, these examples appear to be counter- examples to the above observation in that the degree adverbials apparently co—occur with a compound adjective containing nag that cannot occur with other degree adverbials. But notice that these sentences are unambiguous and are synonymous with (d.i), (d.ii) and (d.iii), respectively: (d) i. kare wa sonnani husyooziki de ya nai. he so dishonest is not 'He is not so dishonest.’ ii. kare wa ammari husyooziki de ya nai. he too dishonest is not 'He is not too dishonest.‘ iii. kare wa motto husyooziki de ya nai. he more dishonest is not 'He is not more dishonest.‘ This observation suggests that the sentences in (0) do not involve the compound adjective husyooziki ma nag "not dishonest" as a unit. That is, they can be analyzed into (e), but not (f): (e) i. kare wa [sonnani husyooziki de] nai. he so dishonest is not ii. kare wa [ammari husyooziki de] nai. he too dishonest is not 164 iii. kare wa [motto husyooziki de] nai he more dishonest is not (f) i. *kare wa sonnani [husyooziki de nai] he so dishonest is not ii. *kare wa ammari [husyooziki de nai] he too dishonest is not iii. *kare wa motto [husyooziki de nai] he more dishonest is not Thus, they are not genuine counter-examples to the above—observed restriction regarding the non-ambiguity of sentences such as (2.28), (2.30), (2.32) and (2.33). On the contrary, the non-ambiguity of the sentences in (c) confirms this restriction. 6. This is equivalent to saying that (2.28), (2.30), (2.32) and (2.33) can be analyzed into (a), but not (b): (a) i. sono siyoonin wa [taihen husyooziki de] nai. the employee _ very dishonest is not ii. sono siyoonin wa [kimi yori husyooziki the employee you more-than dishonest de] nai. is not iii. kareno ie wa [taihen tiisaku] nai. his house very small not iv. kanozyo wa [taihen minikuku] nai. she very ugly not (b) i. *sono siyoonin wa taihen [husyooziki de nai] the employee very dishonest is not 165 ii. *sono siyoonin wa kimi yori [husyooziki the employee you more-than dishonest de nai] is not iii. *kareno ie wa taihen [tiisaku nai] his house very small not iv. *kanozyo wa taihen [minikuku nai] she very ugly not In contrast, ambiguous sentences (2.22) and (2.25) can be analyzed into both (c) and (d): (c) i. sono siyoonin wa [taihen syooziki de] nai. the employee very honest is not ii. sono siyoonin wa [kimi y rl syooziki the employee you more-than honest de] nai. is not (d) i. sono siyoonin wa taihen [syooziki de nai] the employee very honest is not ii. sono siyoonin wa kimi yori [syooziki the employee you more-than honest de nai] is not 7. The form naka is one of the variant forms of the negative naku "not." The relevant variant forms of naku are as follows, with the main environments in which they occur: 10. 166 {-—-—- tta (past particle) ] naka / roo (presumptive particle) {-—--¢ (infinitive form)} naku / te (gerund particle) nake / -—-- reba (provisional particle) nai / --- 0 (present indicative form) For a detailed discussion of ya-attachment, see section 2.6. Incidentally, observe that the Japanese gang is not an exact equivalent of the English any, as will be partly seen from the fact that gang may precede the negative as in (2.73) (Observe the ungrammaticality of the English literal translation of (2.73)). In this connection, consider the following: (a) Ifsono kurasu no dono gakusei mo sono sensei the class in any student that teacher 0 sonkeisita wake]S de wa nai]S respected that is not 'It is not so that every student in the class respected that teacher.’ It should be observed that sentence (a) does not have the meaning that it is not so that any student in the class respected that teacher. These facts demonstrate that the Japanese gang behaves similarly to the English ever , all and the like with respect to negation. This underlying structure involves the assumption given in Kuno (1970) that a contrastive ya is present after 167 a sentence in underlying structure. But even if this assumption turns out to be incorrect, it will not influence our discussion here in an essential way. Furthermore, the structure (2.75) is not exactly the underlying structure for sentences (2.70) and (2.72) but their intermediate structure in that it is derived from the structure (a) by an Optional shift of the quantifier to follow the noun which it modifies: (a) /S\ I NP VP wa / \ | not Quant N NP V past ”sono "" \ I ‘ the minnano gakusei sonkeisuru all student sono respect kurasu no ga that class in sensei 0 teacher Notice that the quantifier minnano "all" drops nn when it follows the noun which it modifies and this is usually the case with quantifiers and numerals in Japanese. A fuller discussion of quantifiers in Japanese will be given in section 2.9.1. It will be sufficient here to mention that the quantifier in (a) is derived by a rule such as: ll. 12. 13. 14. 168 (b) NP'—-9'(S) (Quant) (Det) N Here and elsewhere in this thesis, "synonymous" means "cognitively synonymous," following Chomsky (1965). For instance, Chomsky (1965, 22) says: (9) (i) I expected a specialist to examine John (ii) I expected John to be examined by a specialist The sentences (9i) and (9ii) are "cognitively synonymous": one is true if and only if the other is true. Furthermore, Chomsky (1965, 162) adds as follows: (20) (i) John is easy for us to please - it is easy for us to please John (ii) it was yesterday that he came - he came yesterda In the case of (20), the deep structures of the paired sentences are identical in all respects relevant to semantic interpretation of the sort we are considering here, so that the transformational analysis accounts for the (cognitive) synonymity. See Rosenbaum (1967). Henceforth the adverb "ultimately" will be omitted but the reader should assume that transformations irrele- vant to this thesis may intervene between the rules we discuss and the surface structures of the sentences in the examples. Incidentally, a contrastive ma cannot be attached to taihen ”very," as shown by the ungrammaticality of: (a) *sono siyoonin wa taihen ya syooziki de nai. the employee very honest is not This is due to the idiosyncrasy of taihen "very." 15. 16. 17. 169 When ya is attached to ma, it causes the following change: (a) ma + wa nai - {ma-wa nai Furthermore, the 00pula ma will be derived from ma + amn. To be more exact, therefore, the change in (a) will be something like: (b) aru + wa nai -—9 SAID: (D -wa aru nai ——+ Q.) e wa nai But the point here is that in both (a) and (b) the contrastive ma is attached to the 00pula, deriving ma ya nai and in this sense there is no essential difference between them. Most of the adjectives in Japanese are included in this class and they are typically characterized by the fact that they end with l in their present indicative forms. As regards this underlying structure, we assume, following Fillmore (1968) and Langendoen (1969), that the expletive i; does not exist preceding an embedded sentence in underlying structure, though Rosenbaum's (1967) argument is based on its presence in deep structure. Here, one linguistic fact may be mentioned: in Japanese there is no equivalent of this English ;3 at all. For instance, observe the following: 18. (a) i. ii. iii. ii. iii. 170 That you have kept your promise is known by them. In is known by them that you have kept your promise. ([kimi ga yakusoku o mamotta koto]S wa you promise kept that karera ni sirarete iru]S them. by known is 'That you have kept your promise is known by them.‘ For you to stay here will be impossible. II will be impossible for you to stay here. [[kimi ga kokoni todomaru koto]S wa you here stay for-to hukanoo daroo]S impossible will 'For you to stay here will be impossible.‘ Japanese does not have constructions corresponding to the (ii) sentences involving the expletive i:, but only the counterparts of the (i) sentences, as shown by the (iii) sentences. Kajita (1968) tries to account for the relatedness between the (a) and the (b) sentences in question in terms of a Downgrading transformation that "downgrades" main clauses of a certain type into a constituent of the subordinate clause and at the same time, "upgrades" the original subordinate clause into 19. 20. 171 a new main clause. When M dominates VSoT’ this VS-T must be in the affirmative, as Kajita (1968) correctly points out. For instance, sentences such as (a) are ungrammatical, though (b) is quite grammatical: (a) *Your solution, my does not seem, is rather fig 222° (b) 13 does not seem that your solution is rather .22 £22- Thus, when VS'T is in the negative, the application of Sentence-raising to give (a) must be blocked. Furthermore, we note that the same is true of sentential adjectives. For instance, sentence (0) is ungrammatical if it is to be synonymous with (d): (c) *They will not certainly win the game. (d) It is not certain that they will win the game. This restriction is also applicable to Japanese VS-T and Adjs. To account for this fact, we propose that the structure underlying sentence (2.188c) be as follows: 172 (a) 4 /Sl\ NP VP I I IW/IifikAdv onl: / I\ I /\\, /L\ Adv Det N past Neg V NP N I I I I I I in the orchard A that boy not steal N aplles A fuller discussion of sentences such as (2.188c) will be given in the next chapter. 21. To explain this fact, we hold that the underlying structure for (2.l89c) will be: I) a 1qu/S ]_V\\P Adv /l2w\ I; /NPN\ I /IPN_P\ Det pres Neg th‘at pliyer ncIlt d!) @best. For a detailed discussion of sentences such as (2.l89c), see the next chapter. 22. See footnote 17. 23. The insertion of an seems not to be absolutely 24. 25. 173 necessary in that the following sentence without an may be grammatical and synonymous with sentence (2.199): (a) It is not that many arrows hit the target. In this connection, it should be noted that recent transformational arguments about negation and quanti- fiers (such as cited at the beginning of this chapter) are based on the similarities between negatives and quantifiers. For instance, G. Lakoff (1969) argues that both negatives and quantifiers are generated in the base as verbs of higher sentences, and moreover, as noted in section 1.1, his Quantifier-lowering rule seems to involve not only quantifiers but also negatives. Jackendoff (1969), (1971) also treats negatives and quantifiers similarly when he argues that the surface order of negatives and quantifiers determines the semantic interpretation of a sentence in which they occur. Furthermore, we may mention here that the English incomplete negatives such as fan and little have the property of being both a negative and a quantifier at the same time. After dake "only" is attached to A, giving dake Mary ga, dake must be obligatorily shifted to follow Mary to derive Mary dake ga "only Mary." This shift of nominal adverbials may be compared with the shift of quanti- fiers, for instance, to derive (2.221b) from (2.22la). 26. 27. 28. 29. 174 Based on similar behaviors of numerals to quantifiers, we assume that numerals such as hutatuno "two" may be included in the category of quantifiers. In this connection, see Langacker (1972, 195), who includes many and seven in the same category of quantifiers. As regards post-determiner quantifiers, see Carden (1970a). In particular, noting that post-determiner quantifiers behave differently from pre-determiner quantifiers in a number of ways, he convincingly argues that a post—determiner quantifier comes from a deep-structure non-restrictive relative clause with the quantifier in the overt-predicate position, in a way quite similar to non-restrictive adjectives. For instance, as in the following: Adjective rule‘ .7 (a) The boys, who were many, left. (b) The many boys left. Historically, the article a developed from the numeral an "one" in the following way: an "one" > an > a Incidentally, this diachronic change shows another fact that the article an is an older form than a. Cited from Chomsky (1957:100-101) with a slight change. CHAPTER III IYTERRELATIONS OF NEGATION AND ADVERBIALS Continuing the preceding discussion, we consider, in this chapter, certain classes of adverbials in connection with negation. Our major hypotheses in this chapter are as follows: 1. Nominal adverbials, adverbials of frequency, adverbials of reason and purpose including benefactive adverbials behave like quantifiers with respect to negation, and so they may be treated in a way similar to quantifiers. 2. If 1 is acceptable and established, we may apply the analysis proposed in Chapter II to these adverbials, and by so doing we may test and confirm the validity of this analysis in both Japanese and English. 3. Manner adverbials may not co-occur with verb-phrase negation. If this is acceptable, it will provide further support for the twofold distinction of negation, sentential and verb-phrase, in the proposed analysis. 3.1. Negation and Nominal Adverbials 12 Japanese The first class of adverbials requiring discussion in cOnnection with negation includes dake "only," (de)saemo "even," ma "too, also" and others, which may be called 175 176 nominal adverbials in that they appear attached to noun phrases. For instance, consider the following: (3.1) John dake ga sono himitu o sitteiru. only the secret know '0n1y John knows the secret.‘ (3.2) John desaemo sono himitu o sitteiru. even the secret know 'Even John knows the secret.’ (3.3) John ma sono himitu o sitteiru. also the secret know 'John also knows the secret.’ First of all, let us consider dake "only” with respect to negation: (3.4) [sono syoonen dake ga ohiru o tabe nakattals that boy only lunch eat not-did 'Only that boy did not eat lunch.‘ (3.5) [Isono syoonen dake ga ohiru o tabeta wake]S de that boy only lunch ate that is wa nails not 'It is not so that only that boy ate 1unch.’ Clearly sentences (3.4) and (3.5) are not synonymous. It is observed that in (3.5) nai "not" in the higher sentence includes mama "only" in its sc0pe, and the non-synonymity of (3.4) and (3.5) suggests that gang "only" in (3.4) is outside the scope of negation. This is directly confirmed by the synonymity of (3.4) with (3.6): (3.6) In (3.6) scope of with its (3.4) is 177 [[ohiru o tabe nakatta no]S wa sono syoonen lunch eat not—did that that boy dake da]S only is 'It is only that boy that did not eat lunch.' dake "only" in the higher sentence is outside the negation. The synonymity of (3.4) with (3.6) along non-synonymity with (3.5) demonstrates that dake in outside the s00pe of negation. Furthermore, since the SCOpe of negation is the whole sentence in which it occurs, we maintain that dake in (3.4) is outside the sentence containing the negation in the underlying structure. Thus, the underlying structure for sentence (3.4) will be as follows: (3.7) é//////’ I‘NNIIIIIIIIIIII“Vf /§\‘I‘MM~I\;‘I‘I~P“‘Aux :j:: Ad///D‘et m////3/P \\\\ l I Neg past A sono syoonen ga N taLeru nai that boy I eat not ohiru o /I lunch The structure (3.7) indicates that since the scope of the negative is 32, dake is outside of its SCOpe. Here, note the use of a dummy symbol A which is meant to stand for an 178 unspecified nominal adverbial. The use of A is necessary, in particular, to distinguish the underlying structures for sentences such as (3.8) and (3.9): (3.8) sono gakusei dake ga kono mondai o toka nakatta. that student only this problem solve not-did 'Only that student did not solve this problem.’ (3.9) sono gakusei wa kono mondai dake wa toka nakatta. that student this problem only solve not-did 'Only this problem, that student did not solve.‘ That is, the structures underlying (3.8) and (3.9) will be distinguished in terms of A, as follows, with unnecessary details aside: (3.10) :P/31\VP ./\ \m :11: /IN:\N NPV/P\V\ I only Aivw ‘ M\\\ Neg past A sono gakusei ga Det N toku nai that student I solve not kono mondai 0 this problem (3.11) ’////,»S2‘r--Algya~“u--u‘~‘~‘~ AdVN NP IIIIIIIIII‘7“~V' Aux dike /\ / \ I only Det N NP Neg past gakI / I I sono sei Ava Det N toku nai that student solve not A kono monlai 0 ga this problem The relevant difference between these structures reflects the semantic difference between the sentences derived from them, namely, (3.8) and (3.9). Now, turning to the derivation of sentence (3.4) from (3.7), Sentence-raising applies to (3.7), deriving the intermediate structure: (3.12) 31 NP/ N VP ux //.\\\\ ,/’/7/‘\\\\ I I Ava.th T TP T Nfg past Adi I sono syoonen N taberu nai dake that boy eat not only ohiru 0 ga lunch Then, the adverbial dake "only" must be attached to A. But Iscall that this operation can be covered by the Quantifier- attachment rule in (2.274), repeated here as (3.13): 180 (3.13)(=2.274) {Quant X [A (Det) NJNP Y Adv I—) N Quant E IAva I (Bet) N JNP Y where X and Y are variables, and X X contains no Quant, Ava or Neg This transformation, applied to (3.12), attaches QQEQ "only" to A, giving sentence (3.4).1 Thus, no new transformation is necessary to take care of the derivation of sentences such as (3.4) from (3.7). Here, it may be mentioned that QQEE is not just shifted but attached to the unspecified nominal adverbial A in the sense that sono syoonen dake ga "only that boy" as a unit forms a noun phrase. This can be easily shown by the passive and the reflexive tests as follows: (3.14) kanozyo wa sono syoonen dake ni sukarete iru. she that boy only by liked is 'She is liked by only that boy.‘ (3.15) sono syoonen dake ga zibun(zisin) o kitaete iru. that boy only himself training is ’0nly that boy is training himself.’ Next, let us consider sentence (3.5). Based on the observation that in (3.5) the negative in the higher sentence includes QQEE in its SCOpe, we maintain that the underlying structure for (3.5) will be as follows: 181 (3.16) NP”’//’/”’SI::::::::::::VE‘~‘P‘wa I NP\/ S2\\ux Ieg A nai ////’\;\\\\\‘ N///\\\V not Ava DTt N past dake sono syoonen N taberu only that boy I eat ohiru 0 ga lunch This structure indicates that the nominal adverbial naka "only" is inside the scope of negation. Inserting a comple- mentizer and a copula into 82 and 81’ respectively, we get sentence (3.5). In this particular case, there is no synonymous counterpart of (3.5) which is derived from (3.16) by the application of Contrastive ya-attachment. That is, sentence (3.17) is not synonymous with (3.5) and this is due to the idiosyncrasy of mana "only" in Japanese: (3.17) sono syoonen dake ya ohiru o tabe nakatta. that boy only lunch eat not-did 'Only that boy did not eat lunch.’ In order to derive the underlying structure (3.7), we need the following base rule: (3.18) a.(=2.218a) s ——+ NP VP (Aux) (wa) b. VP —-? Ava In turn, the derivation of the underlying structure (3.16) will require the following rule in addition to (3.18):2 182 (3.19) (s) (AavN> (Det) N} NP ’{s Um) Rule (3.18b) may conjoin with (2.219) into (3.20) and rule (3.19) may conjoin with (2.2180) into (3.21):3 (3.20) (NP) V (Neg) Neg VP'-—9 Quant Ava (3.21) (s) (Ava) (Quant) (Det) N {s (m } The foregoing discussion demonstrates that the semantic difference between (3.4) and (3.5) may be reduced to the difference between sentential and verb-phrase negation: only sentential negation can include nominal adverbials such as gag; "only" in its scope. Adding to this discussion, let us examine next the following sentences involving negation and another nominal adverbial (de)saemo "even": (3.22) Tom desaemo kanozyo o settoku-deki nai. even her persuade can not 'Lit. Even Tom cannot persuade her.‘ (3.23) [[Tom desaemo kanozyo o settoku—dekiru wakeJS even her persuade can that de wa nai]S is not 'Lit. It is not so that even Tom can persuade her.‘ There is some semantic difference between (3.22) and (3.23), though (3.23) may not be used as commonly as (3.22). For 183 4 instance, they will occur in the following contexts: (3.24) kimi ga kanozyo o settoku-deki nai no wa toozen you her persuade can not that no wonder da ga Tom desaemo kanozyo o settoku-deki nai. is but even her persuade can not 'Lit. It is no wonder that you cannot persuade her but even Tom cannot persuade her.‘ (3.25) kimi ga kanozyo o settoku-dekiru no wa toozen you her persuade can that no wonder da ga Tom desaemo kanozyo o settoku-dekiru wake is but even her persuade can that de wa nai. is not 'Lit. It is no wonder that you can persuade her but it is not so that even Tom can persuade her.’ The semantic difference between (3.22) and (3.23) is confirmed by the fact that (3.22) cannot occur in the context of (3.25), nor can (3.23) occur in that of (3.24): (3.26) *kimi ga kanozyo o settoku-dekiru no wa toozen you her persuade can that no wonder da ga Tom desaemo kanozyo o settoku-deki nai. is but even her persuade can not 'Lit. It is no wonder that you can persuade her but even Tom cannot persuade her.’ 184 (3.27) *kimi ga kanozyo o settoku—deki nai no wa toozen you her persuade can not that no wonder da ga Tom desaemo kanozyo o settoku-dekiru wake is but even her persuade can that de wa nai. is not 'Lit. It is no wonder that you cannot persuade her but it is not so that even Tom can persuade her.’ The ungrammaticality of (3.26) and (3.27) shows that sentences (3.22) and (3.23) are not interchangeable in the above contexts; this clearly demonstrates that they are not synonymous. Furthermore, we note that sentence (3.23) may often be replaced by some other sentence. In this connection, compare the following: (3.28) a. Tom desaemo kanozyo o settoku-dekiru. even her persuade can 'Lit. Even Tom can persuade her.‘ b.(=3.22) Tom desaemo kanozyo o settoku—deki nai. even her persuade can not 'Lit. Even Tom cannot persuade her.‘ Sentence (3.28a) implies that there is some other person who gag persuade her, whereas sentence (3.28b) implies that there is some other person who cannot persuade her. Then, re-examine (3.23). Sentence (3.23) in its primary reading means that there is some other person who 9gp persuade her, but Tom cannot persuade her. This is to say that in contrast with someone (who can do so) Tom cannot persuade 185 her. This meaning can be expressed by the following sentence involving a contrastive wa: (3.29) Tom HQ kanozyo o settoku-deki nai. her persuade can not 'Tom (in contrast with someone else) cannot persuade her.’ Sentence (3.29) is more straightforward than (3.23), expressing a similar meaning. To show the semantic simi- larity of (3.29) to (3.23), sentence (3.29) can occur in the context of (3.25) but not of (3.24); hence the grammaticality of (3.30) but the ungrammaticality of (3.31): (3.30) kimi ga kanozyo o settoku-dekiru no wa toozen you her persuade can that no wonder da ga Tom Ea kanozyo o settoku-deki nai. is but her persuade can not 'It is no wonder that you can persuade her but Tom cannot persuade her.‘ (3.31) *kimi ga kanozyo o settoku-deki nai no wa toozen you her persuade can not that no wonder da ga Tom Ea kanozyo o settoku—deki nai. is but her persuade can not 'It is no wonder that you cannot persuade her but Tom cannot persuade her.‘ From the foregoing discussion we see that sentence (3.22) is not synonymous with (3.23). In (3.23) the negative in the higher sentence includes desaemo "even" in its scope, and the non-synonymity of (3.22) with (3.23) suggests that desaemo in (3.22) is not included in the s00pe of the negative. This in turn suggests that since the scope of the negative is the whole sentence in which it occurs, 186 desaemo "even" in (3.22) occurs outside the sentence containing the negative in the underlying structure. Consequently, the structures underlying (3.22) and (3.23) will be (3.32) and (3.33), respectively, with unnecessary details aside: (3032) /S1\ NP VP 1 / S 2\ Ad‘vN NP \VP Aux deslaemo //’\\\ a””””/'l ‘\\\\\\‘ I even ATVN N TP Y Neg pres A Tom‘ga N settoku—dekiru nai persuade can not kanozyo 0 her (3.33) /Sl\ NP \VP wa l NLg \ Aux nai AdézyP ///’ \\\\\ 1 not ‘ TP X pres desaemo Tom ga N settoku-dekiru even I persuade can kanozyo 0 her The relevant difference between them reflects the semantic difference between sentences (3.22) and (3.23). The 187 derivation of (3.22) from (3.32) inVolves the application of Sentence-raising and Quantifier-attachment stated in (3.13). In turn, the derivation of (3.23) from (3.33) requires the insertion of a complementizer and a 00pula into S2 and 81’ respectively. Extending this discussion, we may now consider the third nominal adverbial mg "too, also" as in: (3.34) Mary mg nemura nakatta. either sleep not—did 'Mary did not sleep, either.’ (3.35) Mary mg nemutta wake de wa nai. too slept that is not 'It is not so that Mary slept, too.‘ The nominal adverbial mg behaves very similarly to (de)saemo "even" with respect to negation. This is not surprising after all, since (de)saemo contains m9. It is observed that sentences (3.34) and (3.35) are not synonymous: more Specifically, (3.34) implies that some other person did not sleep either, while (3.35) implies that some other person slept. Thus, (3.35) may be almost synonymous with (3.36) containing a contrastive Ea: (3.36) Mary HQ nemura nakatta. sleep not-did 'Mary (in contrast with someone else) did not sleep.’ The difference between (3.34) and (3.35) may be compared with that between (3.37) and (3.38) in English: (3.37) That pupil was not scolded, either. 188 (3.38) That pupil was not scolded, £99. Sentence (3.37) implies that there was some other pupil who was ngt scolded, whereas (3.38), if it is acceptable, implies that some other pupil was scolded. To put it dif— ferently, sentence (3.38) is synonymous with: (3.39) It is not so that that pupil was scolded, tgg. Summarizing the foregoing discussion, we have demonstrated that nominal adverbials in Japanese show semantic differences, depending upon whether they co-occur with sentential or verb—phrase negation. Clearly this provides further motivation for the twofold distinction of negation, sentential and verb-phrase. 3.2. Negation and Adverbials of Frequency in Japanese Our concern in this section is to show that adverbials of frequency behave similarly to quantifiers with respect to negation. To illustrate the point, let us cite concrete examples such as: (3.40) Mary wa itumo yakusoku o mamora nai. always promise keep not 'Lit. Mary always does not keep her promise.‘ (3.41) [[Mary wa itumo yakusoku o mamoru wake]S de wa always promise keep that is nails not 'It is not so that Mary always keeps her promise.’ Clearly sentence (3.40) is not synonymous with (3.41). We 189 observe first that the negation in (3.41) is sentential and includes in its s00pe the lower sentence containing itumq "always." In contrast, the negation in (3.40) is verb- phrase negation and the non-synonymity of (3.40) with (3.41) suggests that itumg in (3.40) is outside the sc0pe of negation. This is further confirmed by the synonymity of (3.40) with (3.42): (3.42) [[Mary ga yakusoku o mamora nai no]S wa itumg da]S promise keep not that always is 'Lit. It is always that Mary does not keep her promise.’ It is observed that in (3.42) itumg occurs outside the sentence containing nai "not," thus indicating that it is outside the scope of nai. The synonymity of (3.40) with (3.42) and its non- synonymity with (3.41) demonstrate that 133mg in (3.40) is outside the s00pe of negation. Furthermore, the s00pe of negation is the whole sentence in which it occurs. Then, it follows that if 133mg is outside the scope of negation, it must be outside the sentence containing the negation. Based on this consideration, we maintain that the structure underlying (3.40) will be something like: 190 N?””/,,zaSl-1‘\\\\\“~vp I I ”/”””’ 2‘::;1:\\\\\\‘\Aux itE:o ,//’/’/ \\\\\ ) always (3.43) if 1 NP 7 Neg pres Mar ga N mamoru nai keep not yakusoku o promise The structure (3.43) indicates that since the scope of negation is 82, itumo "always" is outside of its s00pe. The derivation of sentence (3.40) from (3.43) requires the 5 application of Sentence-raising. Applying Sentence-raising to (3.43), we derive: (3.44) M//S\1\AVP ux lP /////’| P\\\\\ ‘ ) N NP I NTg pres AdlvF Mar; ga N mamoru nai itumo yakusoku o If the adverbial is shifted to precede the subject noun phrase, the resulting sentence is as follows: (3.45) itumo Mary wa yakusoku o mamora nai. always promise keep not 'Lit. Always LIary does not keep her promise.’ 191 If, on the other hand, itumo "always" in (3.44) is shifted to follow the subject noun phrase, it yields sentence (3.46): (3.46)(=3.40) Mary wa itumo yakusoku o mamora nai. always promise keep not 'Lit. Mary always does not keep her promise.‘ Here we need some transformational rule to take care of the shift of adverbials such as itumg in the derivation of (3.45) and (3.46) from (3.44). This rule, which may be called Adverbial-movement, will be formulated as: (3.47) { NP Advi,} [ X NP Y AdvF JS‘-_9‘X AdvF NP Y where X and Y are variables, and NP is immediately dominated by S6 This transformation is not ad hgg since it is used to shift not only adverbials of frequency but also adverbials of reason and purpose, as will be discussed in the next section. It is now time to consider sentence (3.41). The negation in (3.41) is sentential, as already noted, so sentence (3.41) will have the following underlying structure under our analysis: 192 (3.48) /81\ T? \\\\\\‘VP wa ////S2 N%g NP ‘\\\\\\\\‘VP na1 ; A; v F //’////’\3\\\\\\\‘ 1 NF VP Aux itumo | \\\\ ‘ always N TP V pres Mary ga N mamiru keep yakusoku o promise After S3 is raised into S2, the adverbial itumg "always" is moved to follow the subject noun phrase by the application of rule (3.47). Then, the insertion of a complementizer and a copula into 82 and 81’ respectively, generates sentence (3.41). The relevant difference between adverbials of frequency and quantifiers is that HQ cannot be attached to adverbials of frequency as in the case of quantifiers. In other words, sentences such as (3.49) cannot be synonymous with (3.41): (3.49) lary wa itumo HQ yakusoku o mamora nai. always promise keep not 'Mary usually does not keep her promise (but in this particular case she does).' This is due to the idiosyncrasy of adverbials of frequency in general. In order to derive underlying structures such as 193 (3.43) and (3.48), we will need the following base rules: (3.50) a.(=3.18a) s ——9»NP VP (Aux) (wa) b. VP ——) AdvF Rule (3.50b) may conjoin with (3.20) into: (3.51) {(NP) V (Neg) Neg VP —-9( Quant Adv (Adv N F This rule can also derive structures such as (3.53) underlying sentence (3.52): (3.52) itumo ookuno koohosya ga yakusoku o mamora always many candidate promise keep nakatta. not-did 'Lit. Always many candidates did not keep their promises.‘ (3.53) T? /31\VP l NF///’S2--‘-~“~‘~““-““‘VP Adv Quant ‘F NP/ 3*‘Aux 001kuno / \ / \\ I many itumo always m“- QuTnt T NP \ Nfg past A koohosya ga N mamoru nai candidate keep not yakusoku o promise 194 The structure (3.53) indicates that since the scope of negation is S3, both of the adverbial and the quantifier are outside of its scope. Quite similarly, rule (3.51) along with (3.50a) may give structures such as (3.55) underlying sentence (3.54): (3.54) itumo Mary dake ga yakusoku o mamora nakatta. always only promise keep not-did 'Lit. Always only Mary did not keep her promise.‘ (3.55) NPV/Sl\ / S\VP 11:11:}? itiumo :P JP always NP/ \ AF“ Aux dake I//N?\\ /1P\ only Adv N NP Neg past 111111,! A Mary ga mamoru nai keep not yakusoku o promise This structure shows that the adverbial of frequency and the nominal adverbial are both outside the s00pe of negation. Some other examples containing negation and an adver- bial of frequency may be cited here: (3.56) a. [John wa sibasiba zibunno gimu o hatasa nails often his duty perform not 'John often does not perform his duty.’ 195 b. [[John wa sibasiba zibunno gimu o hatasu often his duty perform wake]S de wa nai]S that is not 'It is not so that John often performs his duty.‘ (3.57) a. [Bill wa taitei sippaisi nai]S im most cases fail not 'In most cases Bill does not fail.’ b. [[Bill wa taitei sippaisuru wake]S de wa nai]S in most cases fail that is not 'It is not so that Bill fails in most cases.’ 3.3. Negation and Adverbials p§.Reason and Adverbials pf Purpose ip Japanese Now, we may proceed to consider the third class of adverbials that behave similarly to quantifiers with respect to negation. Let us take up adverbials of reason first: (3.58) kare wa sore ga riypu d3 syussekisi nakatta. he that reason for attend not-did 'For that reason he did not attend.’ (3.59) [[kare wa sore ga riypu is syussekisita wake]S \ he that reason for attended that de wa nai]S is not 'It is not so that he attended for that reason.’ Obviously sentences (3.58) and (3.59) are not synonymous. We observe that in (3.59) the negative nai in the higher sentence includes the adverbial in its scope, and that the 196 non-synonymity of (3.58) with (3.59) suggests that the adverbial in (3.58) is outside the sc0pe of the negative. This observation is supported by the fact that (3.58) is synonymous with (3.60), in which the adverbial is outside the sentence containing the negative: (3.60) [[kare ga syussekisi nakatta no]S wa sore ga he attend not-did that that riypu gals reason is 'Lit. It is for that reason that he did not attend.’ Since the scope of negation is the whole sentence in which it occurs, the adverbial in (3.60) is outside the scope of negation. Then it follows that the adverbial of (3.58) must be derived outside the sentence containing the negation in the underlying structure. This consideration permits us to _set up the following underlying structure for sentence (3.58): (3.61)TP/81\VP I NP/ SKA? R / \ N Neg past sore ga riyuu de that reason for kare ga syussekisuru nai . he attend not The structure (3.61) shows that since the scope of the 197 negative is 82, the adverbial is outside of its sc0pe. Applying Sentence-raising,7 we derive the intermediate structure: NP VP Aux 1P T 1 Nfg past AdvR kare ga syussekisuru nai s€;e’;a’:iynn\3e Next, applying Adverbial-movement to move the adverbial to follow the subject noun phrase, we get (3.63): (3.63) “ //Dl\\ NP AdvR VP Aux I V (Tg past kare ga sore ga riyuu de syussekisuru nai This structure becomes sentence (3.58). If, on the other hand, the adverbial gppe ga £1133 dg "for that reason" in (3.62) is moved to precede the subject noun phrase kapg ga "he," the resulting sentence is as follows: (3.64) sore ga riypu'dg kare wa syussekisi nakatta. that reason for he attend not-did 'For that reason he did not attend.’ It should be noticed that the movement of the adverbial to derive sentences (3.58) and (3.64) from (3.62) can be taken care of by rule (3.47), if it is slightly revised to: 198 (3.65) {AdvF } Adv AdvF 3 —R P [XNPY(Adv:P]s ’X Adv Y F ) —-— NP AdvR-P where X and Y are variables, and NF is immediately dominated by S This revision makes the rule more general, since it becomes applicable to adverbials of reason and purpose in addition to adverbials of frequency. Turning next to sentence (3.59), it will be derived from the underlying structure (3.66) under our analysis: 3? VP wa NP V‘P nai 1 no S3 AdvR / \\ If? ‘1” “f" N V past sore ga riyuu dé I I that reason for kare ga syussekisuru he attend, This structure indicates that the adverbial is inside the scope of the negative in 81' After 83 is raised into S2, the adverbial is moved to follow the subject noun phrase. Then, inserting a complementizer into 82 and a copula into 81’ we derive sentence (3.59). 199 A similar analysis holds for the following pair of sentences containing an adverbial of reason composed of an embedded sentence: (3.67) [Jane wa sono inu ga kowakatta kara the dog was afraid of because isi o nage nakatta]S stone throw not-did 'Lit. Because Jane was afraid of the dog, she did not throw a stone at it.‘ (3.68) [[Jane wa sono inu ga kowakatta kara the dog was afraid of because isi o nageta wake]S de wa nai]S stone threw that is not 'It is not so that Jane threw a stone at the dog because she was afraid of it.‘ In (3.68), the negation in the higher sentence includes the adverbial in its s00pe and the non-synonymity of (3.67) with (3.68) suggests that the adverbial in (3.67) is outside the sc0pe of negation. This is directly confirmed by the synonymity of (3.67) with (3.69), in which the adverbial is outside the sentence containing the negative: (3.69) [[Jane ga isi o nage nakatta no]S wa [sono inu stone throw not-did that the dog ga kowakatta kara]Adv da] was afraid of because is S 'Lit. It is because Jane was afraid of the dog that she did not throw a stone.’ As the scope of negation is the whole sentence in which it occurs, the adverbial in (3.69) is outside the scope of negation. Accordingly, the underlying structures for (3.67) 200 and (3.68) will be (3.70) and (3.71), respectively, with unnecessary details aside: (3.70) Nfir”””/”S r~“‘-‘1p“\“‘? A AdvR VP 1‘ V Neg past Jane ga sono inu ga the dog _2_% Jane ga N nageru nai / throw no t kowakatta kara isi 0 was afraid of because stone (3.71) NP ’/////’S3 \ i N NP V past ane ga sono inu ga l the dog Jane ga N nageru l throw kowakatta kara isi 0 was afraid of because sons The application of Sentence-raising to (3.70) will derive the following intermediate structure: 201 N NP V Neg past I ll I A Jane ga N nageru nai I throw not isi o Jane ga sono inu ga the dog stone kowakatta kara was afraid of because Next, if Adverbial-movement applies to move the adverbial to follow the subject noun phrase, we get sentence (3.67). If, on the other hand, the same transformation applies in such a way as to move the adverbial in (3.72) to precede the subject noun phrase, the resulting sentence is as follows: (3.73) sono inu ga kowakatta kara Jane wa isi the dog was afraid of because stone 0 nage nakatta. throw not—did 'Because Jane was afraid of the dog, she did not throw a stone at it.‘ In turn, the derivation of sentence (3.68) from (3.71) requires first the raising of 33 into 82. Then, the adverbial is moved to follow the subject noun phrase. Next, the insertion of a complementizer and a 00pula into 32 and 81’ respectively, generates sentence (3.68). The above argument also holds for adverbials of purpose with respect to negation. To cite just one example, consider the following: 202 (3.74) [karera wa kanozyo p tasukeru tameni hontoono they her help to true koto o iwa nakattals thing tell not-did 'They did not tell the truth to help her.‘ (3.75) [[karera wa kanozyo g tasukeru tameni hontoono they her help to true koto o itta wake]S de wa nai]S thing told that is not 'It is not so that they told the truth to help her.‘ Sentence (3.74) may be ambiguous but its primary reading is not synonymous with that of (3.75). In (3.75) the negative in the higher sentence includes the adverbial in its scope, and the non—synonymity of (3.74) with (3.75) suggests that the adverbial in (3.74) is outside the scope of negation. This is clearly supported by the synonymity of (3.74) with (3.76), in which the adverbial is outside the sentence containing the negation: (3.76) [[karera ga hontoono koto o iwa nakatta no]S wa they true thing tell not-did that [kanozyo p tasukeru tame(ni)] da]n AdvP 0 her help to is 'Lit. It is to help her that they did not tell the truth.‘ Based on this observation, we maintain that the underlying structures for sentences (3.74) and (3.75) will be (3.77) and (3.78), respectively: 203 (3.77) s NP VP S Adv ,/”’////’ 2;::\‘\\“~\‘_ P NP VP Aux A//////fi\\\\\\\\\\k l //’/’\:\\\ l karera ga kanozyo o N NP V Neg past they her karera ga hontoono iu nai tasukeru tameni they true tell not help to koto 0 thing (3.78) 1/1\\Neg l///’S ZPPP‘PPPPPNP‘P“-~T NP I Al karera ga kanozyo 6 N V past they her karera ga hontoono kot0'5 iu tasukeru tameni they true thing tell help to Observe the relevant difference between them: in (3.77) the s00pe of the negative is 82 and the adverbial is outside of its SCOpe, whereas in (3.78) the negative, whose s00pe is 81’ includes the adverbial in its s00pe. Now, let us consider the derivation of (3.74) from (3.77). Applying Sentence- raising to (3.77), we get the following intermediate 204 structure: (3.79) S Mk NP VP A P I / \\ 1 I N V Neg past Adv l A ) l /P\ karera ga hontoono kot6' iu nai karera ga kanozyo o o tasukeru tameni Then, the movement of the adverbial gives sentence (3.74). In turn, the derivation of sentence (3.75) from (3.78) requires the raising of S3 into S2 and the movement of the adverbial as well as the insertion of a complementizer and a copula into 82 and 81’ reSpectively. Here, so-called benefactive adverbials may be included, for our purpose, with adverbials of purpose in the broad sense of the word. They behave quite similarly to adverbials of purpose with regard to negation. To illus- trate the point, let us consider the following: (3.80) sensei wa kimino tame p omotte kimi 0 home teacher your good for you praise nakatta. not—did 'For your good the teacher did not praise you.‘ (3.81) [[sensei wa kimino tame p omotte kimi o hometa teacher your good for you praised wake]S de wa nai]S that is not 'It is not so that the teacher praised you for your good.‘ 205 That is, sentence (3.80) is not synonymous with (3.81); only (3.80) can be synonymous with (3.82): (3.82) [[sensei ga kimi 0 home nakatta no]S wa kimino teacher you praise not—did that your tame p omotte da]S good for is 'Lit. It is for your good that the teacher did not praise you.‘ The synonymity of (3.80) with (3.82) suggests that the adverbial is outside the s00pe of negation in (3.80). In (3.81), on the other hand, the adverbial is included in the scope of negation since the negation occurs in a higher sentence than the adverbial. Based on this consideration, we hold that sentences (3.80) and (3.81) will have the following underlying structures, reSpectively: (3.83) s NP Vf Adv /V':2\Aux /P\ [P F ’/,/’3J“\\\\ I kimino tame 6 N NP ‘ Nfg past your good senseilga N homeru nai omotte teacher I praise not for kimi 0 you 206 (3.84) ////,S1....1\~\\\‘\\‘~\~ NP \ VP wa ///’82~\\\\‘\\\\‘ Nrg NP VP nai I I not S Adv NP VP Aux I ///.\\\ I kimino tame o omotte T ?P V past your good for sensei ga N homeru teacher I praise kimi 0 you The structure (3.83) shows that the adverbial is outside the scope of negation, while (3.84) indicates that the negative, whose scope is 31’ includes the adverbial in its SCOpe. Now, the application of Sentence-raising to (3.83) derives the following intermediate structure: (3.85) 31 NP/VP/ \A\VP ux N NP eg past Ava sensei ga I hoLeru nLi I kimino tame o omotte kimi 0 Next, the movement of the adverbial yields sentence (3.80). In turn, the derivation of sentence (3.81) from (3.84) involves the raising of 83 into 82, the movement of the 207 adverbial as well as the insertion of a complementizer and a copula into 82 and 81’ respectively. Now, in order to derive underlying structures such as (3.61), (3.66), (3.70), (3.71), (3.77), (3.78), (3.83) and (3.84), we will need the following base rules: (3.86) a.(=3.50a) S -%’NP VP (Aux) (wa) b. VP —-9.Adv oP Rule (3.86b) may conjoin with (3.51) into: (3.87) (NP) V (Nes)\ Neg I VP "—9 Quant I Ava AdvF AdVRoP I Summarizing the foregoing discussion, we see that adverbials of reason and adverbials of purpose including benefactive adverbials behave quite similarly to quantifiers With respect to negation. In particular, we have noted that they show significant semantic differences, depending upon Whether they occur with sentential or verb-phrase negation. 3.4. The Application pf the Analysis pp the Corresponding Eflgligh Adverbials Our particular concern in this section is to apply the PrOposed analysis of certain classes of Japanese adverbials in regard to negation to the correSponding English adverbials With respect to negation. By so doing, we will attempt to Show its validity and applicability to English as well as to JaPanese. We will consider nominal adverbials, adverbials 208 of frequency and adverbials of reason and purpose in that order. 3.4.1. Negation and Nominal Adverbials 1p Epglish We have called dgkp "only," (de)saemo "even" and g9 "too, also" nominal adverbials in Japanese. Their English equivalents are only, even and £99 or glgg. The arguments given about Japanese nominal adverbials with respect to negatioh are largely valid for English nominal adverbials as well. Let us consider pply first, citing the following examples: (3.88) ‘gply that man does ppp know the secret. (3.89) Np: pply that man knows the secret. We observe that sentence (3.88) is not synonymous with (3.89). We further observe that sentence (3.88) is synony- mous with (3.90), but not (3.91), whereas (3.89) is synonymous with (3.91), not (3.90): (3.90) It is pply that man that does pgp know the secret. ”(3.91) It is p23 so that pply that man knows the secret. It is observed that in (3.90) the adverbial pply, which is outside the sentence containing the negative, is outside of its scope. From the synonymity of (3.88) with (3.90), it follows that pply in (3.88) is outside the sentence containing the negative in the underlying structure. In contrast, the synonymity of (3.89) with (3.91) indicates that not occurs in a higher sentence than only in the 209 underlying structure. Based on this consideration, the underlying structures for sentences (3.88) and (3.89) will be (3.92) and (3.93), reSpectively: (3.92) /91\ NP WP 42 Ava NP/Au/x \VP only Ava Det N pres Neg V NP I I I I I / \ A that man not know Det N the secret (3.93) 31 NP/ \VP Ad’///:it N' pres V NP INII I /\ only that man know Det N the secret The underlying structure (3.92) indicates that pply is outside the sc0pe of p93 in 82, while (3.93) indicates that pply is included in the s00pe of p93 in 81’ Next, observe the use of a dummy symbol A which stands for an unspecified nominal adverbial. The use of A is necessary, in particular, to distinguish the underlying structures of sentences such 210 as (3.94) and (3.95): (3.94) Only Jane does p93 like that man in the group. (3.95) Only that man in the group is not liked by Jane. The underlying structures for (3.94) and (3.95) can be distinguished in terms of A, as follows, with unrelated details aside: (3.96) s / l\ I TP VP I”/,,/’7F2'~\\A\\\\‘N~ AiVN NP Aux VP only I that man in the group 1\ NF VP I NP/AuX/SZ\VP :3: I 1. N/ \ Jale nEf lI dv7702%:::52?3K\\\\\\\\\\\¥ INI I A that man in the group (3.97) s The relevant difference between them is in the position where A occurs; with A removed, (3.96) and (3.97) become identical. Thus, by using A we can distinguish the structures 211 underlying sentences such as (3.94) and (3.95) in a natural way. Now, consider the derivation of sentences (3.88) and (3.89) from (3.92) and (3.93), respectively. Applying Sentence-raising to (3.92), we derive the intermediate structure: (3.98) m%l\\w ,/”'I“\\N I ,////I \\\\\.M IP AivN Det I pres Nng AdlvN A that man not know Det /‘\\\N only the secret Then, Ouantifier-attachment applies, giving sentence (3.88). Note, in particular, that the Quantifier-attachment rule in (3.13), repeated here as (3.99), Can be applied, unchanged, to English as well as to Japanese: (3.99)(=3.l3) Quant Y I X E A (Det) N JNp Adv }.__9 N Quant x [ IAva I (Bet) N 1NP Y where X and Y are variables, and X contains no Quant, Ava or Neg Therefore, the derivation of sentence (3.88) from (3.92) needs no new rule. In turn, consider the derivation of sentence (3.89) from (3.93). First, Sentence-raising applies to (3.93), deriving the intermediate structure: 212 ”‘10” ”N NP Aux VP VP ////'\T\\\\‘ I ///’ \\ ' Ava Det N pres V NP Neg \ l I l / \ I only that man know Det N not I l the secret Next, we need some rule to assign not to only. Here, it should be noticed that Negative-attachment in (2.196) of Chapter II, repeated here as (3.101), can be used for this purpose: X {.Ava Y Neg -—9 AdvF Quant X Neg+ {11.va Y AdvF b. X V Y Neg -9 X Neg+V Y where X and Y are variables, and X contains no Neg, Quant, AdvF or Ava such as gnly The application of this Negative-attachment rule assigns £23 to 9311 in (3.100), giving sentence (3.89). Notice, in particular, that Negative—attachment cannot attach n93 in (3.100) to the verb preceded by ggly to yield sentence (3.88), as follows: (3.102) [[only that man]NP [pres]Aux [know the secret]VP [[nothngVPJSl —9 213 ley that man does ggt know the secret. This derivation is correctly blocked through the filtering function of Negative-attachment. Returning again to (3.93), if Sentence-raising does not apply, it, is and §g are inserted, as in many other cases of Chapter II, generating sentence (3.91). In this case again, no new transformation is necessary to derive sentences (3.89) and (3.91) from the same underlying struc- ture, (3.93), nor is a new rule needed to block the derivation of sentence (3.88) from the structure (3.93). In order to derive underlying structures such as (3.92) and (3.93), we will need the following base rules: (3.103) a.(=2.279a) S -9'NP (Aux) VP b. V]? ——)Ava c. NP ——-) (Ava) (Det) N (S) Rule (3.103b) may conjoin with (2.280) into (3.l04a), and (3.l03c) may conjoin with (2.2790) into (3.104b):9 (3.104) a. (Neg) V (NP) Neg Ava b. (Ava) (Quant) (Det) N (S) NP—aI(NP)S } Taking this argument one step further, we may proceed to consider examples such as the following: (3.105) a. The student did g9: solve ggly that problem. b. iny that problem was n9: solved by the student. 214 c. £21.2flii that problem was solved by the student. We observe that sentence (3.105a) is synonymous with (3.105c), but not with (3.105b). According to our approach, (3.105a) and (3.105c) will be derived from the same underlying structure: (3.106) [[[[the student]NP [past]Aux [[solve]V [ggly If Passivization does not apply to (3.106), Sentence—raising and Negative-attachment apply to assign £2: to the verb, generating sentence (3.105a). If, on the other hand, Passivization applies to (3.106), followed by the application of Sentence-raising, it will derive the intermediate structure (3.107): (3.107) [Eggly that problem]NP was solved by the student [[nOtJNngVPJSl Observing the structure (3.107), we note that Negative- attachment (3.101a) can apply to attach n93 to gnly, deriving sentence (3.1050). But Negative—attachment (3.101b) cannot apply to (3.107) to assign n91 to the verb, since the verb is preceded by gnly. Thus, the following derivation is blocked: (3.108) [[only that problem]NP was solved by the student [EEQEJNngVPJS '7%9 1 Only that problem was not solved by the student. The foregoing discussion demonstrates that no new 215 transformation is necessary to account for the synonymity of (3.105a) with (3.1050). Turning next to sentence (3.105b), it will be derived from the underlying structure (3.109) with unnecessary details aside: (3.109) [[[[the student]NP [past]Aux [[QQEJNeg [solve]V This structure indicates that ggly in S1 is outside the scope of n23, thus explaining why p93 in (3.105b) cannot include ggly in its scope. If no moving transformation, including Passivization, applies to (3.109), Quantifier- attachment is blocked since A follows p93, thus giving no sentence. In this way, through the filtering function of this transformation we can block the derivation of sentence (3.105a) from (3.109). If, on the other hand, Passivization applies to (3.109), the resulting structure is as follows: (3.110) [[[[A that problem]NP was not solved by the Then, Quantifier-attachment can apply, following the appli- cation of Sentence-raising, to assign gnly to A, deriving sentence (3.105b). Thus, this approach can explain why (3.1050), but not (3.105b), is the passive counterpart of (3.105a) derived from the same underlying structure. Furthermore, this approach correctly blocks the derivation of (3.105b) from the structure underlying (3.105a) and (3.1050) as well as that of (3.105a) from the structure underlying (3.105b). 216 Amplifying this discussion, consider next the following examples: (3.111) a. Not ggly that girl hit Bill. b. ley that girl did ggt hit Bill. 0. Bill was n93 hit by ggly that girl. Sentence (3.1110) is synonymous with (3.111a), but not with (3.111b), and this fact can be explained in a similar way. According to our analysis, sentences (3.111a) and (3.1110) will be derived from the same underlying structure: (3.112) [[[[[onlylAva [that]Det [girleJNP [past]Aux [hit BillJVPJS 1m, [rnotheglVPJS 2 ‘ 1 Applying Sentence-raising and Negative-attachment to assign n9: to gnly, we derive sentence (3.111a). If, on the other hand, Passivization applies to (3.112), the following structure is derived: (3.113) [[[Bill was hit by only that girljs ]NP 2 r LrnOtJNngVPJSl Then, Negative-attachment attaches ggt to the verb, after 32 is raised into 81’ yielding sentence (3.1110). Thus, (3.111a) and (3.1110) are derived from the same underlying structure, which accounts for the synonymity of these sentences. Sentence (3.111b) is derived from: (3.114) [[[[[AlAdv [that1Det [girlINJNP [past]Aux N [[notJNeg hit 131111VPJS2JNP [[onlyJAvaJVPJSl After the application of Sentence—raising, 217 Quantifier-attachment applies to attach only to A, yielding sentence (3.111b). If, on the other hand, Passivization applies to (3.114), it derives the intermediate structure: (3.115) [[[Bill was not hit by [A that girlJNPJSZJNP Then, Quantifier-attachment is blocked since A follows the negative n91. This provides an explanation for the lack of a passive counterpart of (3.111b), derived from the underlying structure (3.114). Turning next to another nominal adverbial gygg, let us compare the following sentences: (3.116) N91 gygn that man knows the secret. (3.117) Ezgn that man does pg: know the secret. The English gygn does not seem to behave like ggly with respect to negation. That is to say, sentences (3.116) and (3.117) seem to be synonymous with each other. Also, compare them with: (3.118) It is ppt so that gggn that man knows the secret. Sentence (3.118) appears to be synonymous with both (3.116) and (3.117). Furthermore, consider the following: (3.119) The secret is n93 known gzgg to that man. Again, sentence (3.119) may be synonymous with (3.118) as well as (3.116) and (3.117). These observations suggest that the sentential hierarchy of gygn and a negative is not relevant to the meaning of the sentence in which they occur. Recall, in this connection, that the Japanese equivalent of even, namely, (de)saemo, behaves similarly to the English 218 qnly with respect to negation: the sentential hierarchy of (de)saemo and negation is relevant to the meaning of the sentence in which they occur. This seems to have some relation to the fact that (de)saemo contains mg "also." The next nominal adverbial to be considered here is tgg. First, compare the following: (3.120) The boy did 222 kiss Mary, :99. (3.121) The boy did 22: kiss Mary, either.10 If sentence (3.120) is acceptable, there is some semantic difference observed between (3.120) and (3.121): (3.120) implies that the boy kissed someone else, while (3.121) implies that there was someone else in question whom the boy did not kiss. In other words, sentence (3.120), but not (3.121), is synonymous with (3.122): (3.122) It is not so that the boy kissed Mary, 102- Therefore, sentences (3.120) and (3.122) may be derived from the same underlying structure distinct from that underlying sentence (3.121). In particular, the negation in (3.120) and (3.122) is sentential, while that in (3.121) is verb- phrase negation. 3.4.2. Negation and Adverbials 9; Frequency in English The above argument regarding Japanese adverbials of frequency with respect to negation may hold for English adverbials of frequency as well. To take a concrete example, let us compare the following pair of sentences: (3.123) That player often does not do his best. 219 (3.124) That player does not often do his best. We observe that sentences (3.123) and (3.124) are not synonymous. To clarify this point, we may cite the following: (3.125) It is o_f_t_en that that player does 119:0. do his best. (3.126) It is .1121 so that that player m does his best. We observe that (3.123) is synonymous with (3.125), not (3.126), whereas (3.124) is synonymous with (3.126), not (3.125). In (3.125) gftgg is outside the s00pe of negation, Since it is outside the sentence containing the negation. Then, the synonymity of (3.123) with (3.125) suggests that Qiggn in (3.123) is outside the scope of negation. The Scope of negation is the whole sentence in which it occurs, 30 if gftgn is outside the scope of negation, it must be Outside the sentence containing the negation in the uIlderlying structure. In contrast, in (3.126) pp: in the higher sentence includes the adverbial 9.23.9.9. in its 800pe, arid the synonymity of (3.124) with (3.126) permits us to aSSUme that M in (3.124) occurs in a higher sentence than m in the underlying structure. This consideration will lTad us to set up (3.127) and (3.128) as the underlying StI‘uctures for (3.123) and (3.124), reSpectively, with 1111related details aside: 220 (3.127) ’///,//81 NP VP ' A! v F ar””//’7§2~\N\\\“\\ ‘ NP Aux VP often \\\\ 1 ,/’//7 ‘\\\\\ Det pres Nfg Y NP tiat player not do fiis best (3.128) /sl {P VP .L N‘ 2 9g NP WP not ’/////,§? Ad|vF Aux VP often Det N ples K ‘\\\ that player do his best The structure (3.127) indicates that the adverbial is outside the scope of the negative in S2, whereas (3.128) shows that the negative, whose scope is 81’ includes the adverbial in its 800pe. Now, let us consider the derivation of sentence (3.123) from (3.127). The application of Sentence-raising moves S up into S deriving the 2 1’ intermediate structure: 221 (3.129) s //’ 1N NP Aux VP VP ///"\\\ 1 ,/’//;Z \\\\\ I Det T pres Neg I N? AivF tAat player nLt do his bes often The adverbial gftgg in (3.129) is shifted to follow the subject noun phrase,11 yielding sentence (3.123). If, on the other hand, the adverbial is moved to precede the subject noun phrase, the resulting sentence is as follows: (3.130) gftpn that player does pot do his best. This movement of gitgn may be compared with the movement of Japanese adverbials of frequency to follow or precede the subject noun phrase, as seen in section 3.2. To take care of the movement of gftgn in the derivation of (3.123) and (3.130) from (3.129), we need a transformation such as:12 (3.131) [ X NP Aux Y AdvF JS-—-9 NP Aux Adv M Fh Adv NP Aux F where X and Y are variables, and NP is immediately dominated by S It should be noted that this transformation is very similar to the corresponding Japanese transformation in (3.47): the relevant difference is the presence of Ag; in (3.131) and its absence in (3.47). Thus, rule (3.131) may conjoin with (3.47) into: 222 (3.132) [ X NP (Aux) Y AdvF ls NP (Aux) Adv A { F} Y AdvF NP (Aux) —-a where X and Y are variables, and NP is immediately dominated by S Then, rule (3.132) is applicable to English as well as to Japanese. The applicability of the same transformation to English and Japanese adverbials of frequency reduces the ad hgg-ness of this transformation. Furthermore, just as the same transformation can be used for adverbials of reason and purpose including benefactive adverbials in Japanese, so it can be used for the corresponding English adverbials as well, as will be discussed in the subsequent section. These facts will sufficiently show the independent motivation of this transformation in English grammar. Turning back to (3.127), if Sentence-raising does not apply, that, it and lg are inserted, and the embedded sentence is extraposed, yielding sentence (3.125). In this way, we need no new rule except Adverbial-movement (3.131) to derive (3.123), (3.125) and (3.130) from the same underlying structure, (3.127), nor do we need a new rule to block the derivation of (3.124) from (3.127).13 Now, consider the derivation of (3.124) from (3.128). After 83 is raised into 82 and the adverbial is moved, 82 is further raised into 81’ giving the following intermediate structure: 223 (3.133) ,,,,’v”’::::::;Ti:::::::::“‘--~11~_ NP Aux AdvF VP VP / \ l I \ I Det N pres often V NP Neg L \ A l t at player do hlS best not Then, Negative-attachment stated in (3.101) applies to assign the negative not to often, giving (3.134): (3.134) S NP Aux AdvF ‘///VP\\\ Dét//‘\\\N pies Nég/ NP V tAat plager nLt often go figg:;;§¥ Next, the auxiliary d9 is introduced,14 yielding sentence (3.124). It should be mentioned here that through the filtering function of Negative-attachment we can block the attachment of not to the verb, as in the following: . r 1 . (3.135) t[that player]NP [pres]Aux [ofteniAdv [do hlS F best]VP [[MjNngVPJSl '7‘9 That player giggg does p91 do his best. Thus, no new transformation is necessary to derive sentence (3.124) from (3.128), nor is a new transformation necessary to block the derivation of sentence (3.123) from (3.128). If, on the other hand, Sentence-raising does not apply to (3.128), that, 13, lg and g9 are inserted, as in similar cases discussed in Chapter II, to generate sentence (3.126). No new transformation is needed in this case either. 224 In order to derive the underlying structures (3.127) and (3.128), we need the following base rules: (3.136) a.(=3.103a) s -+'NP (Aux) VP b. VP —-) AdvF Rule (3.136b) may conjoin with (3.104a) into: (3.137) (Nee) v (NPM Neg ‘ VP -9 Quant ) Ava AdvF 1 It should be mentioned that rules (3.136a) and (3.137) can also derive the structures underlying sentences such as (3.138) and (3.139): (3.138) iny that player gfpgn does pgt do his best. (3.139) 'Many players gitgn do pgt do their best. Their underlying structures will be (3.140) and (3.141), respectively: NP V I f 82 \ Ava TP/// fiP only / SB\\ AdlvF Aux VP often AKD‘et N prl‘es Neg/ \V\NP N 1 ‘ ‘ I ’ f§:::::> A that player not do 18 best 225 (3.141) /8 1\VP NP VPI / 2\ Q‘Tm NP many /:3\VP AleF NP Aux often Quant N pres Neg I I I IV A players not do their best These structures indicate, among others, that only, often and many are outside the s00pe of negation, based on the constraint that the scope of negation is the whole sentence in which it occurs. Other related examples may be cited here: (3.142) a. 19 most cases Mary does 999 keep her word. b. Mary does 999 keep her word 99 most cases. b’. It is 999 so that Mary keeps her word l9 most cases. (3.143) a. My brother usually does 999 use knife and fork. b. My brother does not usually use knife and fork. b’. It is 999 so that my brother usually use knife and fork. (3.144) a. It is always not good to live alone. b. It is not always good to live alone. b’. It is 999 so that it is always good to live alone. 226 3.4.3. Negation and Adverbials 99 Reason and Adverbials 99 Purpose 99 English The above discussion concerning Japanese adverbials of reason and of purpose with respect to negation is also applicable to the corresponding English adverbials. First of all, let us consider adverbials of reason, citing the following example: (3.145) Tom did 999 adopt the plan for that reason. Sentence (3.145) may be ambiguous with two readings, depending upon whether the adverbial for that reason is outside the scope of negation or not. When the adverbial is included in the s00pe of negation, it is synonymous with: (3.146) It is 999 so that Tom adopted the plan :9; that reason. Yet, if for that reason is outside the s00pe of negation, it is synonymous with the following: (3.147) a. For that reason Tom did 999 adopt the plan. b. It was for that reason that Tom did 999 adOpt the plan. Based on these observations, we maintain that (3.145) is derived from either of the following underlying structures, depending on its reading: 227 VP ( 3.148) s NP/ l\ I I 82 AdvR ”/”’i:::; ‘\\\\\\‘ Z//h\\\\\\‘\\\i NP A VP l TX //”//?/ \\\\\\ for that reaSEfi N past Neg V /NP\ Tom not ad0pt Det N the plan (3.149) I l / 2\ eg NP NP Jot S Adv 3 R //\ NP Aux VP I I ./l ‘\\SN for that reasdfi N past V P l I /\ Tom ad0pt Det N the plan The structure (3.148) shows that since the scope of the negative is 82, the adverbial is outside of its scope, whereas the structure (3.149) indicates that the negative, whose scope is S1, includes the adverbial in its s00pe. Now, taking up (3.148) first, the application of Sentence-raising gives (3.150): 228 (3.150) m% l Aux VP P . I /I\. I Adv I I I /\ R Tom not adopt Det N l for that reasEn t e plan The structure (3.150) becomes sentence (3.145). Then, the 15 Optional shift of for that reason gives sentence (3.147a). Furthermore, if S2 in (3.148) is not raised into S that, 99 1’ and 999 will be inserted to generate sentence (3.147b). In turn, let us consider the derivation involving (3.149). First, the application of Sentence-raising raises S into S . Then, if S 3 2 2 1’ and 99 are inserted, as in other similar cases, deriving is not raised into S that, 99, 99 sentence (3.146). 0n the other hand, if S2 is raised into S Negative-attachment assigns not to the verb, yielding l’ sentence (3.145). In particular, Negative-attachment must block the attachment of 999 to the verb, if the adverbial 999 9999 reason precedes the verb. For instance, after 83 is raised into S2 in (3.149), the adverbial may be Optionally shifted to precede the subject noun phrase, deriving: 229 (3.151) u——?’ *6 <: *0 Adv/ me nLt R I I / \ N past V NP for that reasofi Tim adlpt Défr \\\N I the plan After the raising of S2 into 81’ Negative—attachment must be blocked so that it may not perform the following derivation: (3.152) [[for that reason) Tom past ad0pt the AdvR plan [[notJNngNPJS'vf9 For that reason Tom did not adOpt the plan. To block this derivation, Negative—attachment in (3.101) requires only a slight revision as follows: (3.153) a.(=3.101a) Quant X {Ava} Y Neg—-) AdvF Quant X Neg+ {Ava Y AdvF b.(=3.lOlb) X Y Y Neg—9 X Neg+V Y where X and Y are variables, and X contains no Neg, Quant, AdvF, Ava such as only or AdvR-P Next, let us consider examples such as the following: 230 (3.154) Your father did 999 scold you because 99 999 999. Sentence (3.154) may be similarly ambiguous with two readings. More specifically, it can be synonymous with either of the following: (3.155) It is 999 so that your father scolded you because 99,999 999. (3.156) a. Because your father was sad, he did 999 scold you. b. It was because your father was sad that he did 999 scold you. In (3.155) 999 in the higher sentence includes the adverbial in its scope, while in (3.156) the adverbial is outside the scope of negation, particularly in (3.156b), the adverbial is outside the sentence containing the negation. Consequently, we hold that sentence (3.154) is derived from either of the following underlying structures, depending upon its reading: (3.157) [[[[your father]NP [past]Aux [[999JNeg [scold]V [youJNPJVPJSZJNP [[because your father was figdlAdvRJVPJSl (3.158) [[[[[Eyour father]NP [past] [[scold]V Aux [youJNPJVPJS JNP [[because your 9ather was §§91Adv JVPJS JNP [EEQEJNngVPJS R 2 1 The derivation of (3.154) from (3.157) is rather straightfor- ward: the application of Sentence-raising generates sentence (3.154). Next, the Optional movement of the adver- bial gives sentence (3.156s). In contrast, the derivation of sentence (3.154) from (3.158) involves the application of 231 Sentence-raising and Negative-attachment. Incidentally, notice that if the adverbial in (3.158) is shifted to precede the subject noun phrase, followed by the application of Sentence-raising, it derives the following intermediate structure: (3.159) [[because your father was sad]AdVR your father past scold you [[nOtJNngVPJS 1 Then, the revised Negative-attachment rule in (3.153) correctly blocks the following derivation: (3.160) [[because y9ur father was sad]Adv your R father past scold you [[notJNngVPJS -f9 1 Because your father was sad, he did 999 scold you. Thus, Negative—attachment can block the derivation of sentence (3.156a) from (3.158) just as it blocks the genera- tion of (3.147s) from (3.149). The above argument also holds for adverbials of purpose with regard to negation. To illustrate with an example, consider the following: (3.161) She did 999 attend the meeting 99 entertain those pe0p99. We Observe that sentence (3.161) may be ambiguous with two readings.l6 More specifically, it is synonymous with either of the following: (3.162) It is 999 so that she attended the meeting 99 entertain those peop99. (3.163) a. 99 entertain those peOple she did 999 attend the meeting. 232 b. It was 99 entertain those people that she did 999 attend the meeting. In (3.162) the negation in the higher sentence includes the adverbial in its s00pe, while in (3.163) the adverbial is outside the scope of negation. Accordingly, (3.161) may be derived from either of the following structures with minor details aside, depending upon its reading: (30164) /Sl\ NP VP I I ,/’//:;;§2‘\\\‘\\\ AdVP NP Aux VP Z//\\\\X I I ’//’/1/ \\\\\\ to entertaifi N past Nfg N NP those people sAe not attend Det N I I the meeting (3.165) 31 TF////’ ‘~\~\\\\\\\~Vf 82 Neg NP’/777’, 777777777‘VP nLt I I NP’/j;gf%5‘\\\\VP A/figzg7w777“‘-llyy“_ /\ to entertain those peOple NE\\\' she attend Det N I t e meeting 233 The derivation of sentence (3.161) from (3.164) involves the application of Sentence-raising. Next, the optional movement of the adverbial yields sentence (3.163a). In turn, the derivation of (3.161) from (3.165) requires the application Of Negative-attachment as well as Sentence-raising. In particular, if the adverbial is optionally shifted in the 82-0ycle of (3.165) and 82 is raised into S the 1’ intermediate structure (3.166) is derived: (3.166) N past v NP Neg to entertain those I I //’\\\\ I people she attend Det N not the meeting Then, Negative-attachment cannot apply to attach.999 to the verb, since the verb is preceded by the adverbial: that is, the following derivation is blocked: (3.167) [[99 entertain those peOple]Ava she past attend the meeting [[nOtJNngVPJS -f9 1 99 entertain those people she did 999 attend the meeting. Thus, the generation of (3.163a) from (3.165) is correctly blocked through the filtering function of this transformation. The above argument can be extended to benefactive adverbials with respect to negation. To cite just one 234 example, observe the following: (3.168) I did 999 do it for her sake. Sentence (3.168) may be ambiguous and be synonymous with either of the following: (3.169) It is 999 so that I did it for her sake. (3.170) a. For her sake I did 999 do it. b. It was for her sake that I did 999 do it. In (3.169) the negation in the higher sentence includes the adverbial in its scope, while in (3.170) the adverbial is outside the scope of negation, especially in (3.170b), the adverbial is outside the sentence containing the negation. Thus, we hold that sentence (3.168) may be derived from either of the following structures, depending on its reading: (3.171) [[[[IJNP [past]Aux [[9991Neg IdOJV [itJNPJVPJS2JNP [[for her sakelAvaJVPJSl (3.172) [IIIIIIJNP [past]Aux [[dolv [itJNPJVPJSBJNP [[for her sakeJAvaJVPJSZJNP [[QQEJNngVPJSl Applying Sentence-raising to (3.171), we derive sentence (3.168). Then, the Optional movement of the adverbial gives sentence (3.170a). In turn, the derivation of (3.168) from (3.172) requires the application of Sentence-raising and Negative-attachment. Furthermore, if the adverbial in (3.172) is optionally moved to precede the subject noun phrase, the resulting structure is: (3.173) [[[[for her sake]Adv I past do it]S JNP P 2 [Egg—JNngVPJSl 235 Next, S is raised up into 81’ but the subsequent applica- 2 tion of Negative—attachment is blocked; not cannot be attached to the verb preceded by the adverbial, as in the following: (3.174) [[for her sake]AdVP I past do it [InotINngVPISl .99. For her sake I did 999 do it. In this way, the derivation of (3.170a) from (3.172) is correctly blocked by the filtering function of Negative-attachment. Now, in order to derive the underlying structures (3.148), (3.149), (3.157), (3.158), (3.164), (3.165), (3.171) and (3.172), we will need the following base rules: (3.175) a.(=3.136a) s —->NP (Aux) VP b. VP -)Adv R°P Rule (3.175b) may conjoin with (3.l37) into: (3.176) {(Neg) V (NP)( Neg Quant N (AdvF ) AdvR-P Summarizing the foregoing discussion, we have demonstrated that adverbials of reason and adverbials of purpose including benefactive adverbials in English also behave similarly to quantifiers with respect to negation: they show semantic differences depending on whether they co-occur with sentential or verb—phrase negation. This 236 constitutes additional support for the distinction between sentential and verb-phrase negation and also shows the validity of the analysis proposed in Chapter II. 3.5. Negation and Manner Adverbials Somewhat different from those adverbials previously discussed, manner adverbials do not occur with both types of negation, sentential and verb-phrase: they may co-occur with one of them. Nevertheless, because there is still a sig- nificant similarity of manner adverbials to the above-cited adverbials as well as to quantifiers with respect to negation, we need to discuss them in this section. 3.5.1. Negation and Manner Adverbials 99 Japanese In section 2.2 we have briefly considered manner adverbials in connection with negation in Japanese, citing the following example: (3.177)(=2.40) kanozyo wa koohukuni sina nakatta. she happily die not-did 'She did not die happily.‘ It has been noted that sentence (3.177) is synonymous with (3.178) in which the negative in a higher sentence includes the manner adverbial in its s00pe: (3.178)(=2.4l) [[kanozyo wa koohukuni sinda wakels she happily died that de wa nai]S is not 'It is not so that she died happily.’ 237 At first sight, sentence (3.177) seems to pose no relevant problem, especially when we take into consideration the fact that the scope of negation is the whole sentence in which it occurs. To clarify the point, let us compare (3.177) with the following: (3.179) kare wa kinoo sono siken o uke nakatta. he yesterday the examination take not—did 'He did not take the examination yesterday.’ (3.180) boku wa tosyokan 99 sawaga nakatta. I library in make a noise not-did 'I did not make a noise in the library.‘ In both (3.179) and (3.180) the negative includes in its SCOpe the adverbial kinoo ”yesterday" and tosyokan 99 "in the library," respectively, so they are synonymous with (3.181) and (3.182), respectively: (3.181) [[kare wa kinoo sono siken O uketa wake]S he yesterday the examination took that de wa nai]S is not 'It is not so that he took the examination yesterday.’ (3.182) [Iboku wa tosyokan 99 sawaida wake]S de I library in made a noise that is wa nai]S not 'It is not so that I made a noise in the library.‘ The comparison of (3.177) with (3.179) and (3.180) presents no apparent differences between them. On closer examination, however, a certain difference can be observed: sentence 238 (3.177) does not mean that she did 999 die, while sentences (3.179) and (3.180) mean that he did 999 take the examina- tion and that I did 999 make a noise, respectively. This difference is made clearer by reference to the ungrammati- cality of (3.183) and the grammaticality of (3.184) and (3.185): (3.183) *kanozyo wa koohukuni sina nakatta, sunawati she happily die not-did that is kanozyo wa sina nakatta. she die not-did 'She did not die happily, that is, she did not die.‘ (3.184) kare wa kinoo sono siken o uke nakatta, he yesterday the examination take not-did sunawati kare wa sore o uke nakatta. that is he it take not-did 'He did not take the examination yesterday, that is, he did not take it.‘ (3.185) boku wa tosyokan 99 sawaga nakatta, I library in make a noise not—did sunawati boku wa sawaga nakatta. that is I make a noise not-did 'I did not make a noise in the library, that is, I did not make a noise.‘ Clearly this difference must be accounted for somehow in a Japanese grammar, if the grammar is to be descriptively adequate. In the present framework, this difference can be explained in terms of the distinction between sentential and 17 verb-phrase negation. That is to say, our proposal is as in the following: 239 (3.186) The negation in sentence (3.177) originates from sentential negation, while that in (3.179) and (3.180) is verb-phrase negation. To support this proposal, the following can be cited: A. The negative in (3.177) cannot combine with the verb in any context. If it does, the resulting sentence is ungram- matical as follows: (3.187) *[Ekanozyo ga sina-nakatta no]S wa koohukuni she die not-did that happily da)S 1s 'Lit. It is happily that she did not die.‘ As contrasted with (3.187), both of the following are quite grammatical: (3.188) [[kare ga sono siken o uke-nakatta nOJS he the examination take not-did that wa kinoo da]S yesterday is 'Lit. It is yesterday that he did not take the examination.‘ (3.189) [[boku ga sawaga-nakatta no]S wa tosyokan ' I make a noise not-did that library (99) daJS in is 'Lit. It is in the library that I did not make a noise.‘ In (3.187) the adverbial koohukuni "happily" in the higher sentence includes the negation in its scope, and the ungrammaticality of (3.187) supports the observation in 240 section 2.2 that manner adverbials cannot include negation in their s00pe. 18 B. The difference observed in A between sentential and verb-phrase negation is also observed in those sentences with negation and quantifiers, discussed in Chapter II. For instance, let us reexamine the following examples from this viewpoint: (3.190) a. subeteno oobosya ga sintaikensa 0 all applicant physical-examination uke nakatta. undergo not-did 'All the applicants did not undergo a physical-examination.‘ oobosya ga subete sintaikensa o applicant all physical-examination uke nakatta. undergo not—did 'All the applicants did not undergo a physical-examination.' (3.19l)(=2.140) oobosya ga subete 99 sintaikensa O applicant all physical-examination uke nakatta. undergo not-did 'Not all the applicants underwent a physical-examination.‘ The negation in (3.190) is verb—phrase negation, while that in (3.191) is sentential negation, as discussed in section 2.7. Sentence (3.191) does not mean, though (3.190) means, that the applicants did not undergo a physical-examination. Accordingly, (3.190), but not 241 (3.191), is synonymous with (3.192): (3.192) [[sintaikensa o uke nakatta no]S wa physical-examination undergo not-did that subeteno oobosya da]S all applicant is 'Lit. It is all the applicants that did not undergo a physical-examination.‘ The non-synonymity of (3.191) with (3.192) indicates that sentential negation in (3.191) cannot negate the verb alone. Thus, although in (3.191) the negative is apparently combined with the verb, it does not negate the verb itself but the whole sentence, as is demonstrated by the synonymity of (3.191) with (3.193) involving sentential negation: (3.193)(=2.14l) [Isubeteno oobosya ga sintaikensa all applicant physical-examination o uketa wake]S de wa nai]S underwent that is not 'It is not so that all the applicants underwent a physical—examination.‘ The comparison between (3.190) and (3.191) demonstrates that sentential negation cannot negate the verb alone even when the negation is directly combined with the verb, as in (3.191), as Opposed to verb-phrase negation that is not only combined directly with the verb but also negates the verb itself, as in (3.190). It is important to note a striking parallel between (3.191) and (3.177). Sentence (3.191) cannot have the interpretation in which the nega- tive negates the verb alone. This is exactly the case with 242 (3.177); as shown by the ungrammaticality of (3.183) and (3.187), sentence (3.177) cannot have the interpretation in which the negative negates the verb alone. This consid— eration further supports the proposal in (3.186). C. In comparison with (3.177), examine the following: (3.194) kanozyo wa koohukunimo sina nakatta. she happily die not-did 'She, happily, did not die.‘ The only difference between (3.177) and (3.194) is the minor morphological one of koohukuni "happily as a manner adverbial" versus koohukunimo "happily as a sentential adverbial," and the latter presumably analyzes into koohukuni + 99, though it is interpreted as a single formative unit. This minor difference causes a clear semantic difference between (3.177) and (3.194); (3.194) means, though (3.177) does not mean, that she did not die. Thus, as contrasted with the ungrammaticality of (3.183), (3.195) is quite grammatical: (3.195) kanozyo wa koohukunimo sina nakatta, sunawati she happily die not-did that is kanozyo wa sina nakatta. she die not—did 'She, happily, did not die, that is, she did not die.’ If the negation in (3.177) is sentential negation, that in (3.194) is verb-phrase negation. This is further confirmed by the fact that sentence (3.194) is not synonymous with (3.196) involving sentential negation: 243 (3.196) ?[[kanozyo wa koohukunimo sinda wake]S de wa she happily died that is nails not 'It is not so that she, happily, died.’ Apart from the acceptability of (3.196), sentence (3.196) is not synonymous with (3.194). This difference between (3.194) and (3.177) is quite parallel to that between verb-phrase negation and sentential negation observed in those sentences discussed in sections 2.6, 3.1, 3.2 and 3.3. These observations will be sufficient to support and establish the proposal (3.186). Then, we can prOpose another putative universal by the use of the twofold distinction of negation--verb-phrase negation in this case: (3.197) Manner adverbials cannot co-occur with verb-phrase negation. Returning to sentence (3.177), the foregoing discus- sion demonstrates that its underlying structure will be as follows: (3.198) 31 ,/’/////’ 7777\‘\~ NP VP a I I 82 Neg PP \VP\A 1' \ ux na1 A AdV//’ \\\\V pAst not I I M I kanozyo ga koohukuni sinu she happily die Applying Sentence-raising and Contrastive 99-attachment in (2.114) of Chapter II to attach 99 to the adverbial or to the verb modified by the adverbial, we derive either of the following intermediate structures: NP/ ISIP \\AuxVP A AdV/// \\\\\V p::t J:g I I M\ I kanozyo ga koohukuni wa Slnu nai bONP///Sl\ux I /\ p)... N Adv NLg I I”"' I\ I kanozyo ga koohukuni sinu wa nai (3.l99) a. VP Then, Negative-attachment applies to (3.199), giving the following sentences: 245 (3.200) a. kanozyo wa koohukuni 99 sina nakatta. she happily die not-did 'She did not die happily.’ b. kanozyo wa koohukuni sini 99 si nakatta. she happily die do not—did 'She did not die happily.‘ In the case of (3.200a), the 99 may be optionally deleted, giving sentence (3.177). This 99 deletion cannot apply to (3.200b) since the resulting sentence is ungrammatical as follows: (3.201) *kanozyo wa koohukuni sini si nakatta. she happily die do not-did In turn, if S2 of (3.198) is not raised up into S the l? complementizer wake "that" and the 00pula 99 "is" are inserted into 82 and 81’ reSpectively, as in the derivation of the sentences discussed in section 2.6, yielding sentence (3.178). In order to derive the underlying structure (3.198), we will need the following base rules: (3.202) a.(=3.86a) S ——>NP VP (Aux) (wa) b. VP ——9 (Ade) (NP) V Rule (3.202b) may conjoin with (3.87) into: (3.203) {(Ade) (NP) V (Neg)\ Neg Quant I W “'M Ava I AdvF J IAdvR.P The above discussion demonstrates that no new 246 transformation is necessary to derive sentences (3.177), (3.178), (3.200a) and (3.200b) from the structure (3.198). Furthermore, this analysis can account for a number of above-noted different behaviors of manner adverbials from time and location adverbials with respect to negation. In sum, the analysis proposed in Chapter II has its validity further confirmed when applied to manner adverbials in connection with negation. 3.5.2. Negation and Manner Adverbials 99 English The foregoing discussion regarding manner adverbials in Japanese is largely true of English manner adverbials as well. To start with, consider the following: (3.204) a. Bill did 999 answer wisely. b. It is 999 so that Bill answered wisely. We observe that sentence (3.204a) is synonymous with (3.204b). So, apparently the relation of (3.204a) and (3.204b) seems similar to that of (3.205a) and (3.205b) or that of (3.206a) and (3.206b): (3.205) a. Bill did 999 answer yesterday. b. It is not so that Bill answered yesterday. (3.206) a. Bill did not answer 99 the class. b. It is not so that Bill answered 99 the class. A closer examination of them, however, shows some differences between (3.204) and (3.205) or (3.206). First, compare the ungrammaticality of (3.207) with the grammaticality of (3.208) and (3.209): 247 (3.207) *Bill did 999 answer wisely, that is, he did not answer. (3.208) Bill did 999 answer yesterday, that is, he did not answer. (3.209) Bill did 999 answer 99 the class, that is, he did not answer. Second, observe the ungrammaticality of (3.210) and the grammaticality of (3.211) and (3.212): (3.210) *It was wisely that Bill did 999 answer. (3.211) It was yesterday that Bill did 999 answer. (3.212) It was 99 the class that Bill did 999 answer. In addition, the manner adverbial wisely in (3.204) cannot be preposed jumping over the negative. If it is so preposed, the resulting sentence is ungrammatical as in (3.213) or the adverbial ceases to be a manner adverbial as in (3.214): (3.213) *How wisely Bill did 999 answer! (3.214) Wisely Bill did 999 answer. By comparison, yesterday and 99 the class may be so preposed without affecting the semantic contents of (3.205) and (3.206): (3.215) Yesterday Bill did 999 answer. (3.216) 99 the class Bill did 999 answer. Sentence (3.214) is not synonymous with (3.204), whereas (3.215) and (3.216) may be synonymous with (3.205) and (3.206), respectively. Comparing (3.214) with (3.204), a difference is observed in the surface order of 999 and wisely: in (3.204) wisely follows the negative, while in 248 (3.214) wisely precedes the negative. Moreover, we note that wisely in (3.214) is not a manner adverbial any more but a sentential adverbial. This difference of the surface order of 999 and wisely corresponds to the slight morpho- logical difference between kenmeini "wisely as a manner adverbial" and kenmeinimo "wisely as a sentential adverbial" in Japanese, as noted in connection with koohukuni "happily as a manner adverbial" versus koohukunimo "happily as a sentential adverbial" in section 3.5.1. In English, manner adverbials in general cannot precede a negative, as shown by the ungrammaticality of (3.213) and the non—synonymity of (3.214) with (3.204), and this is exactly the reflection of the observation made in 3.5.1: manner adverbials cannot include negation in their s00pe. These observations together suggest that the negation in (3.204) is different from that in (3.205) and (3.206). In our analysis, if the negation in (3.205) and (3.206) is verb-phrase negation, that in (3.204) is sentential negation. In particular, the ungrammaticality of (3.207) and (3.210) indicates that the negative in (3.204) cannot combine with the verb alone, as Opposed to the negative in (3.205) and (3.206) that can negate the verb alone, as is shown by the grammaticality of (3.208), (3.209), (3.211) and (3.212). Furthermore, if the negation in (3.204) is sentential, as proposed here, this provides further support for the putative universal stated in (3.197), namely: manner adverbials cannot co-occur with 19 verb-phrase negation. 249 The preceding argument leads us to set up (3.217) as the structure underlying both (3.204a) and (3.204b): /S]_\ N VP I, II 2 eg Aux VP (3.217) NP ‘ ’//// \\\ not N past V Ade I Bill answer wisely The application of Sentence-raising and Negative-attachment to assigh not to the verb yields sentence (3.204a). In particular, suppose wisely is Optionally shifted to precede the verb as in (3.218b): r . (3.218) a. LIIIBlllJNP [past]Aux [[answer]V [the question that perplexed everyone]NP lIVPJSZJNP [EnglengVPJSl -—+ b. [[[[BillINP [pastJAux ffiflléél13ldv [wiselyIAdvD [answer] M V [the question that perplexed GVGTYOHGJIWJVPJSZJIW [[nOtJNGgJVPJSl Then, after S2 is raised into 81’ Negative-attachment may attach 999 to the adverbial to derive sentence (3.219), but not to the verb following the adverbial to derive (3.220): (3.219) Bill did not wisely answer the question that perplexed everyone. 250 (3.220) Bill wisely did not answer the question that perplexed everyone. Thus, Negative-attachment in (3.153) will have to be slightly revised to: (3.221) a. Quant ' Ava X AdvF Y Neg -9 Adv Quant Ava X Neg+ AdvF Y Ade b.(=3.153b) X V Y Neg-—-$ X Neg+V Y where X and Y are variables, and X contains no Neg, Quant, Adv , Ava such as 999y, AdvR.P or 9MA The revised Negative-attachment rule can correctly block the following derivation: (3.222) [Bill past wisely answer the question that perplexed everyone [[EQEJNngVPJS "f9 Bill wisely did 999 answer the question that perplexed everyone. Returning to (3.217), if 82 is not raised into 81’ that, 99, 99 and 99 will be inserted, as in the derivation of similar sentences discussed in Chapter II, to derive sentence (3.204b). Thus, no new transformation is necessary to derive sentences (3.204a) and (3.204b) from the same underlying 251 structure, (3.217), nor is a new transformation necessary to block the derivation of non-synonymous sentences. In order to derive the underlying structure (3.217), we will need the following base rules: (3.223) a.(=3.l75a) s -—9’NP (Aux) VP b. VP -—9'V (NP) (Ade) Rule (3.223b) may conjoin with (3.176) into: (3.224) (Neg) V (NP) (Ade) Neg Quant VP 5 Ava AdvF AdvR-P 3.6. Conclusions In this chapter we have discussed four classes of adverbials with respect to negation in both Japanese and English. These discussions demonstrate that: 1. Nominal adverbials, adverbials of frequency and adver— bials of reason and purpose including benefactive adverbials behave similarly to quantifiers with respect to negation in that, depending upon whether they co-occur with sentential or verb-phrase negation, they show semantic differences. This constitutes another motivation for the twofold distinction of negation, sentential versus verb-phrase. 2. The validity of the analysis in terms of Sentence- raising, Negative-attachment and Quantifier-attachment 252 is further confirmed in this chapter, since these rules are useful in accounting for a number of facts concerning the interrelations of negation and those adverbials discussed above. Both 1 and 2 are Observed in both Japanese and English. This fact lends additional confirmation to the proposed analysis. The discussion of manner adverbials in regard to negation provides further support for the proposed analysis and for the twofold distinction of negation. Moreover, we have proposed a presumably universal constraint that is, at least, applicable to both Japanese and English: manner adverbials cannot co-occur with verb-phrase negation. The discussion of these adverbials with respect to nega- tion requires a slight revision of Negative-attachment, Quantifier-attachment and Sentence-raising. This revision makes these transformations more general in that they are made applicable to the derivation of sentences involving these adverbials and negation. Finally, the base rules stated in section 2.12 must be further expanded so that they may cover these adverbials. They are as follows: Japanese: English: 253 S ——-> NP VP (Aux) (wa) (Ade) (NP) V (Neg)\ Neg Quant I VP ’ .Ava I AdvF ) AdVR-P (s) (Ava) (Quant) (Det) N 3..., NP (Aux) VP (Neg) V (NP) (Ade)\ Neg I Quant I VP ’ Ava Adv Advi.P } (Ava) (Quant) (Det) N (s) NP‘—’(INP)S I CHAPTER III FOOTNOTES After dake "only" is attached to A, giving dake sono syoonen g9, 9999 must be obligatorily shifted to follow sono syoonen "that boy" to derive sono syoonen dake 99 "only that boy." This shift may be treated in a way similar to that of quantifiers to derive, for instance, (b) from (a): (a)(=2.225a) sono kurasu no subeteno gakusei ga the class in all student sono sensei o sonkeisi nakatta. that teacher respect not-did 'All the students in the class did not respect that teacher.’ (b)(=2.225b) sono kurasu no gakusei ga subete sono the class in student all that sensei O sonkeisi nakatta. teacher respect not—did 'All the students in the class did not respect that teacher.‘ The relevant difference is that the shift of quantifier subete "all" is Optional, as shown by the grammaticality of (a), and this is usually the case with quantifiers. But in the case of some quantifiers, the shift is obligatory. For instance, consider the following: 254 255 (c)(=2.224) *sono kurasu no minnano gakusei ga sono the class in all student that sensei o sonkeisi nakatta. teacher reSpect not-did 'All the students in the class did not respect that teacher.‘ (d)(=2.213) sono kurasu no gakusei ga minna sono the class in student all that sensei o sonkeisi nakatta. teacher reSpect not-did 'All the students in the class did not respect that teacher.‘ The ungrammaticality of (0) indicates that the shift of minna "all" is obligatory as in the case of nominal adverbials. With the use of rule (3.19) we can distinguish a semantic difference, for instance, between (a) and (b): (a) (b) sono syoozyo dake ga kare O kiratteiru. that girl only him dislike '0nly that girl dislikes him.‘ sono syoozyo wa kare dake o kiratteiru. that girl him only dislike 'That girl dislikes only him.’ The semantic difference between them is due to the dif- ference in the position where dake "only" occurs. This is correctly predicted by the difference of their underlying structures as follows: 256 (C) [[[dakejAva [sonOJDet [syoozyo gaJNJNP only that girl [[kare O]NP [kiratteirulvlVP [preSJAuxls him dislike (d) [[[sonoJDet [syoozyo saJNINP [IiiakelAdv that girl only N [kare OJNJNP [kiratteirulV]VP [preSJAuXJS him dislike Furthermore, we note that rule (3.19) derives underlying structures for sentences such as (e): (e) ?sono syoonen dake ga kanozyo dake o sitteiru. that boy only her only know 'Lit. Only that boy knows only her.‘ If this sentence is acceptable (which seems to be the case with some native speakers of Japanese), then no problem occurs for the base rule (3.19). But sentence (e) sounds unacceptable to some other native speakers including me. If so, the derivation of sentences like (e) must be blocked in some way. This may be done perhaps in terms of a constraint such as: (f) Nominal adverbials may not occur more than once in a simplex sentence. But consider next the following sentence: (g) sono syoonen dake ga kanozyo saemo sitteiru. that boy only her even know 'Only that boy knows even her.‘ Sentence (g) sounds acceptable though it involves two occurrences of nominal adverbials dake "only" and saemo 257 "even." Thus the constraint (f) is too strong and needs to be revised to (h): (h) The same nominal adverbial may not occur more than once in a simplex sentence. With this constraint along with the base rule (3.19) we can generate grammatical sentences involving nominal adverbials and also block the generation Of unacceptable sentences such as (e). Needless to say, constraint (h) is unnecessary for the grammar of the speakers who accept sentences such as (e). 0n the other hand, if one attempts to take care of the semantic difference between sentence (a) and (b) by means of a transformation, such as Kuroda's (1969b) Attachment transformation, this transformation obviously is not meaning-preserving. This analysis has a clear disadvantage: to give up the meaning-preserving condition on transformations is to increase the descrip- tive power of transformations, which in turn contributes to making more serious the defect of the theory of transformational grammar (see Chomsky (1972:124-125)). Rule (3.20) along with (3.21) can also derive structures such as (a) (underlying sentence (b)): 258 NP VP ////S2__9‘~“*“-“-“-~ Quant NP VP subeteno I I all ,r””/"85:::::::::‘~“““‘-.~ AiVN NP > VP Aux dake // \\ \:\\\\\ I only (Quant N NP V Neg past I I /I\ I I A gakusei Ava Det N toku nai student I I solve not A sono mondai 0 ga that problem (b) subeteno gakusei ga sono mondai dake wa all student that problem only toka nakatta. solve not-did 'Only that problem, all the students did not solve.’ Note the two occurrences of A in (a). Moreover, notice that there is no need for the Specific distinction of A for an unSpecified nominal adverbial from A for an unspecified quantifier, because they are dominated by distinct constituents, Ava and Quant, in underlying structure. 4, If the literal English translation for both (3.24) and (3.25) sounds unacceptable, this is due to the difference Of even from saemo with respect to negation, as will be 259 discussed in section 3.4.1. ES. If Sentence-raising does not apply to (3.43), the complementizer 99 "that" and the 00pula 99 "is" as well as the tOpic 99 are inserted, giving sentence (3.42). 6.. Notice that Adverbial-movement follows Sentence-raising in application. For instance, it applies to (b) which is derived from (a) by the preceding application of Sentence-raising: NP VP 2 Adv / \\_ F NP V Aux itumo I /////IR\\\\\ I always N NP N Neg pres Marylga N mamoru nai keep not yakusoku o promise NP /VP\\Aux NP N NP % Neg pies AdvF Mary ga N mamoru nLi idumo yakusoku 0 Thus, the NP in question is immediately dominated by the 260 81' In general, a subject NP is immediately dominated by S. If 82 of (3.61) is not raised up into 31’ "is," the complementizer 99 "that" and the tOpic—marking the 00pula 99 99 will be inserted, deriving sentence (3.60). Incidentally, notice that the attachment of 999y to A in (3.97) is blocked since A follows 999, But if Passivi- zation applies to (3.97), it derives the following structure: (a) [[[A that man in the group is not liked by Jane] 1 [[onl J 1 1 82 NP ——1 Ava VP 81 Then only may be attached to A, after the raising of 82 into 81’ giving sentence (3.95). Rule (3.104a) along with (3.104b) can also derive underlying structures such as (a) (underlying sentence (b)): 261 Np’/////’SI777777777‘7777‘VP l A v VP///’82777777777777777"VP I N I only I I //// 3::N;;\77777777“VP m::: /N\ I /‘I\ I (a) Quant N could Neg VP I I I /\ A applicants not satisfy Ava N A John (b) Q99y John, 999y applicants could 999 satisfy. Notice that there is no need for the specific distinction of A for an unspecified nominal adverbial from A for an unspecified quantifier, since they are dominated by different constituents in underlying structure. When Topicalization applies to 33’ it will derive the intermediate structure: (0) [[[[[IIAJAva John]NP [IA] applicants]NP Quant could not satisfy1831NP [[mgngQuantJVPJSZJNP [EQEAXJAdVNJVPJSl After S is raised into 82, Quantifier-attachment 3 assigns many to A. Then, S2 is further raised into 81’ and only is attached to A, generating sentence (b). Moreover, observe the following sentence: (d) Not only that boy did not observe the rule. 262 If this sentence is acceptable, it is derived from the underlying structure (8), which is also derived by the rules (3.104a) and (3.104b): NP/Sl\ VP I I IP/ 2\ VP I (e) N not S Adv 3 N \A\V I NP only /I\ //VPNP\ Ad N Det N past Neg III I I /NPN\ A that boy not observe Det the rule The structure (e) also underlies sentence (f): (f) It is not so that only that boy did not Observe the rule. 10. We ignore here, as irrelevant, the fact that both (3.120) and (3.121) may be ambiguous. Sentence (3.121) may occur, for instance, in both of the following: (a) Bill did not kiss Mary and the boy did not kiss 999y, either. (b) In addition to refusing to kiss Helen, 999 99y did not kiss Mayy, either. 263 11. To be more exact, often is shifted to follow the auxiliary since adverbials of frequency like often usually follow auxiliaries (see footnote 12): (a) often That player willialways do his best. usually Thus the shift of often in (3.129) to follow the auxiliary derives the following structure: //31\ NP ATx A (b) VP IVF ,//// \\\\\ pres often Neg I N that player' not do his best Then the tense pres is attached to the VP, giving: (0) 9,//”’::::;?SI\\\\\‘- NP Ad‘vF /////;;?VN.\\\\\\ Often Aux Neg N NP that playef' pres not do his best Next, 99 is inserted by what Klima (1964, 256) calls 99—support (such as Tense--?'do + Tense), giving: IVP/IW/‘/81\VP OI; /'\\. Neg V NP /m\ II that plays? do pres not do his beéf' (d) 264 The structure (d) becomes sentence (3.123). The tense—attachment transformation to derive structures such as (c) from (b) will be formulated as: (e) i. pres Ipast Y [ Neg V Z JVP ——9 pres XY [I past} .Neg V Z JVP 11. presN X (past Y E V A JVP —’ pres pres IpastI Y [ V+ Ipast} Z JVP where X, Y and Z are variables, and Y contains no Neg or NP This transformation has no relation to Klima's Tense- attachment (see Klima (1964, 256)). Needless to say, the 99—support transformation must follow the appli- cation of rule (e.i). 0n the other hand, rule (e.ii) is applicable when a verb phrase contains no negative as in the following: (f) [that player]NP [pres]Aux [Often]AdvF [do his best]VP —-9 [that player]NP [often]AdVF (-$)That player often does his best.) [do+pres his best]VP Moreover, it should be noted that rule (e) is in any case necessary to generate ordinary sentences in the present or past tense such as: (g) i. That player does his best. 265 ii. That player did his best. The generation of sentence (g.i), for instance, goes through the following derivation: NP Aux I /VP\ pres V NP I that playEP' do his best 11. Nfi,”””/’7'S\\\\VP V’/// \\\\NP that playEP do pres his best Note that the derivation of (h.ii) from (h.i) is taken care of by rule (e.ii) with Y being null. 12. It should be noticed that in rule (3.131) 999F is shifted to follow 999 rather than the subject-NP, so that (3.131) is general enough to cover derivations such as: (a) [[that player]NP [99991Aux [do his [[OftenlAdvFJVP bestJVP [[alwayslAdvFJVP JS'-9 [[usuallyJAdVFIVP often That player will.{always do his best. usually 13. 14. 15. 266 Notice that the transformation (3.131) correctly blocks the following derivation: (a) [[that player]NP [pres]Aux [not do his best]VP [EQEEEEJAdvFIVPJS"*’ That player pres not often do his best. (--9 (3.124) That player does not often do his best.) Thus the derivation of (3.124) from (3.127) is blocked by the filtering function of (3.131). It should be noted that the tense-attachment transfor- mation (e) in footnote 11 cannot apply to (3.134); the presence of 999 outside the VP in (3.134) does not satisfy the condition ”Y contains no Neg" of this transformation. Then the 99—support transformation applies to (3.134), deriving: NP /Au< /AdvF /VP\IP do pres Neg V that player not often dL his best The structure (a) becomes sentence (3.124). Notice that the shift of adverbials of reason and purpose can be taken care Of by the Adverbial-movement rule in (3.132), if it is slightly revised to: 16. 17. 267 (a) AdvF [ X NP (Aux) Y {AdvR-P} JS '—’ NP (Aux) AdvF (m. } ( )I Y NP Aux AdvR-P where X and Y are variables, and NP is immediately dominated by S To be more exact, sentence (3.161) is three-ways ambiguous and its third reading is that she did not attend the meeting which was held as an entertainment for those people. But we ignore the third reading as irrelevant to our discussion here. One might argue that manner adverbials are more closely tied to the verb than time and location adverbials, and that this causes their differences noted in this section, by supposing that the negative in (3.177) negates the verb and the manner adverbial as a unit. But this argument is not very convincing when we consider examples such as: (a) kanozyo wa [[zyoozuni]AdVM [sono uta OJNP she well the song [utattalvlvP sang 'She sang the song well.‘ Those who make this argument will not deny that an Object noun phrase is more closely tied to a verb than a manner adverbial in that a manner adverbial modifies 268 a unit of an Object noun phrase and a verb. Next, Observe the following: (b) kanozyo wa sono uta o utawa nakatta. she the song sing not—did 'She did not sing the song.‘ (0) kanozyo wa zyoozuni utawa nakatta. she well sing not-did 'She did not sing well.’ The negation in (b) may negate the verb alone, as is seen from the fact-that (b) may be synonymous with: (d) [[kanozyo ga utawa nakatta no]S wa sono uta da]S she sing not-did that the song is 'Lit. It is the song that she did not sing.‘ This is further confirmed by the grammaticality of (e): (e) kanozyo wa sono uta o utawa nakatta, sunawati she the song sing not—did that is (uta o) utawa nakatta. song sing not-did 'She did not sing the song, that is, she did not sing.‘ Compare (d) and (e) with (f) and (g), respectively: (f) *[[kanozyo ga utawa nakatta nOJS wa zyoozuni da]S she sing not—did that well is 'Lit. It is well that she did not sing.’ (g) *kanozyo wa zyoozuni utawa nakatta, sunawati she well sing not-did that is (uta o) utawa nakatta. song sing not-did 'Lit. She did not sing well, that is, she did not sing.‘ 269 The ungrammaticality of (f) and (g) indicates that the negative in (0) cannot negate the verb alone. Furthermore, consider sentence (h): (h) kanozyo wa zyoozuni sono uta o utawa nakatta. she well the song sing not-did 'She did not sing the song well.’ Compare the grammaticality of (i) with the ungrammati- cality of (j) and (k): (i) [[kanozyo ga zyoozuni utawa nakatta no]S wa she well sing not-did that sono uta da]S the song is 'Lit. It is the song that she did not sing well.‘ (j) *[[kanozyo ga sono uta o utawa nakatta no]S wa she the song sing not-did that zyoozuni da]S well is 'Lit. It is well that she did not sing the song.‘ (k) *kanozyo wa zyoozuni sono uta o utawa nakatta, she well the song sing not-did sunawati sono uta o utawa nakatta. that is the song sing not-did 'Lit. She did not sing the song well, that is, she did not sing the song.’ If the ungrammaticality of (f) and (g) is explained by supposing that a manner adverbial is so closely connected with a verb that a negative cannot negate a verb alone but a unit of a verb and a manner adverbial, then there is no reason why (d), (e) and (i) are not 18. 270 ungrammatical: an object noun phrase is more closely related to a verb than a manner adverbial, so a negative should negate a unit of a verb and an object noun phrase, but not a verb alone. Yet, as a matter of fact, as the grammaticality of (d), (e) and (1) shows, a negative can negate a verb alone, in addition to negating a unit of a verb and an Object noun phrase. From this consideration we must conclude that the above assumption based on the close relation of manner adverbials to a verb cannot give a satisfactory expla- nation for the behavior of manner adverbials with respect to negation. We may consider here the reason why a manner adverbial cannot include negation in its s00pe. We note first that a manner adverbial can be associated only with verbs (excluding stative verbs) but not with adjectives, adverbs etc.. More specifically, it can modify only a concrete action described by a verb or verb phrase, and in this sense it can include a verb or verb phrase in its s00pe: (a) [I x JVP Ade l where X contains a verb with the feature [- stative] This very nature or function of manner adverbials has much to do with their inability to include negation in their scope. Suppose, for example, a manner adverbial 19. 271 includes negation in its s00pe: (b) [I Neg x JVP Ade I where X contains a verb with the feature [- stative] Then, it follows that manner adverbials can modify a negated action that has no reference at all: [ Neg X JVP describes no concrete action at all, as "not answer," for instance, refers to no action in answering. In other words, (b) says that manner adver- bials can modify the way an action is 999 performed. But how can we describe the way an action is not performed if the action is not performed at all? Thus it seems to be a linguistic universal that a manner adverbial cannot include negation in its s00pe, since it is based on the factual knowledge that we cannot describe the way an action is not performed. There seem to be some counter-examples to this constraint. For instance, observe the following: (a) She reluctantly did 999 answer the telephone. If sentence (a) is acceptable, this is clearly a counter-example to this constraint in that the manner adverbial reluctantly co-occurs with verb-phrase nega- tion. Semantically, "not answer" in (a) refers to an action such as "refrain from answering" rather than no action at all. Thus sentence (a) will be semantically close to sentence (b): 272 (b) She reluctantly refrained from answering the telephone. The same is true of the following examples: (0) He hesitantly did 999 accept the offer. (d) He regretfully did 999 accept the offer. That is to say, they are semantically close to (e) and (f), respectively: (e) He hesitantly refused the offer. (f) He regretfully refused the offer. Thus, it seems to be the case that certain manner adverbials like reluctantly, hesitantly and regretfully can co-occur with negated verbs which have non- negative equivalents such as: (g) i. not answer —- refrain from answering ii. not accept -—- refuse Another matter to consider is that these manner adver- bials have some negative element of meaning. In any case, the fact still remains that sentences such as (a), (c) and (d) are counter— examples to this constraint. If there are a consid- erable number of counter—examples like them, we will have to revise this constraint so that it may cover those counter-examples in some way. At present the author cannot offer a satisfactory solution to this problem, partly because he cannot find a sufficient number of counter-examples of this kind. Incidentally, note that sentences such as (a), 273 (c) and (d) are not counter-examples to our analysis regarding negation and adverbials. The discussion in this chapter demonstrates that only manner adverbials cannot co-occur with verb—phrase negation. This will be summarized as: (d) Adver- bials Adv Adv Adv Adv Negation N F R P M Sentential X X X X Verb-phrase X X X Counter—examples such as (a), (c) and (d) exactly serve to fill up this gap. Viewed from another standpoint, we may say that the grammars of the speakers who find sentences like (a), (c) and (d) acceptable are undergoing a kind of simplification in that they are generalizing the occurrence of adverbials with reSpect to both types of negation and further limiting the exception. Furthermore, it should be noted that there is a limit to how far such simplification can proceed. CHAPTER IV NEGATIVE RAISING Negative-raising or Negative tranSportationl is discussed by both generative and interpretive semanticians, such as R. Lakoff (1969a), G. Lakoff (l970a), Lindholm (1969), and Jackendoff (1971). Generative semanticians like G. Lakoff and R. Lakoff prOpose Negative-raising as a syn- tactic rule, while interpretive semanticians like Jackendoff reject this approach but try to account for a number of facts regarding Negative—raising in terms of a semantic interpre- tation rule. It is not our concern here to support or deny Negative-raising as a syntactic rule, but to demonstrate that our proposed analysis may take care of a number of facts concerning Negative-raising without using this rule. To be more Specific, our proposal is that if we adOpt the analysis proposed in Chapter II, we may dispense with a minor rule such as Negative-raising in both Japanese and English grammars. If this proposal becomes established, it provides another support for the analysis proposed in Chapter II, in addition to eliminating the minor rule of Negative-raising. 274 275 4.1. Negative-Raising Versus the Proposed Analysis 12 Japanese If we follow R. Lakoff (1969a), Lindholm (1969), G. Lakoff (1970a) and others, and apply their analysis to Japanese, sentence (4.2) will be derived from the structure underlying (4.1) by the application of Negative—raising: (4.1) (watasi wa) John ga Mary 0 syootaisi nai to omou. I invite not that think 'I think that John will not invite Mary.‘ (4.2) (watasi wa) John ga Mary 0 syootaisuru to I invite that omowa nai.2 think not 'I don't think that John will invite Mary.‘ But, as they admit, Negative—raising is a minor rule in that it is applicable to a small class of verbs, namely, non- factive verbs of mental action such as phipk, suppose, believe, gpggg and gap: in English or their equivalents in Japanese. Our concern here is to explore an analysis that may account for the synonymity of (4.1) and (4.2) without using a minor rule such as Negative-raising. To start with, consider the following examples and compare them with (4.1) and (4.2): (4.3) (watasi wa) John ga Mary 0 syootaisuru no de wa I invite that is nai to omou. not that think 'I think that it is not so that John will invite Mary.’ 276 (4.4) (watasi wa) John ga Mary 0 syootaisuru to ya I invite that omowa nai. think not 'I don't think that John will invite Mary.‘ We observe that both (4.3) and (4.4) are synonymous with (4.1) and (4.2). By synonymity we mean that sentences (4.1), (4.2), (4.3) and (4.4) are synonymous in the sense that active and passive sentences are synonymous. To clarify the point, observe the following: (4.5) John wa Mary 0 aisa nakatta. love not-did 'John did not love Mary.' (4.6) Mary wa John ni aisare nakatta. by loved not—was 'Mary was not loved by John.’ First, the synonymity of (4.5) and (4.6) permits us to conjoin them in terms of sunawati "that is" as follows: (4.7) a. John wa Mary 0 aisa nakatta, sunawati Mary wa love not-did that is John ni aisare nakatta. by loved not-was 'John did not love Mary, that is, Mary was not loved by John.‘ b. Mary wa John ni aisare nakatta, sunawati John by loved not-was that is wa Mary 0 aisa nakatta. love not-did 'Mary was not loved by John, that is, John did not love Mary.‘ 277 We observe that both (4.7a) and (4.7b) are grammatical. Next, if we negate one of (4.5) and (4.6), and conjoin it with the other in terms of sunawati "that is," we get an ungrammatical sentence in either case: (4.8) a. *John wa Mary 0 aisita, sunawati Mary wa John loved that is ni aisare nakatta. by loved not-was 'John loved Mary, that is, Mary was not loved by John.‘ b. *Mary wa John ni aisareta, sunawati John wa by was loved that is Mary 0 aisa nakatta. love not-did 'Mary was loved by John, that is, John did not love Mary.’ Clearly both (4.8a) and (4.8b) are ungrammatical or seman- tically anomalous. Third, any of the following sentences can occur following both (4.5) and (4.6): (4.9) a. sikasi Bill wa Mary 0 aisita but loved 'but Bill loved Mary' a’. sikasi Mary wa Bill ni aisareta but by was loved 'but Mary was loved by Bill' b. sikasi John wa Jane 0 aisita but loved 'but John loved Jane' bl I C 278 . sikasi Jane wa John ni aisareta but by was loved 'but Jane was loved by John' sikasi John wa Mary to kekkonsita but married 'but John married Mary' . sikasi Mary wa John to kekkonsita but married 'but Mary married John' In this sense, active and passive sentences are synonymous with each other. Our next step is, then, to show that in a similar sense (4.1) and (4.2) are synonymous. First, let us conjoin (4.1) and (4.2) in terms of sunawati "that is" as in: (4.10) (4.11) (watasi wa) John ga Mary 0 syootaisi nai to I invite not that omou, sunawati (watasi wa) John ga Mary 0 think that is I syootaisuru to omowa nai. invite that think not 'I think that John will not invite Mary, that is, I don't think that John will invite Mary.‘ (watasi wa) John ga Mary 0 syootaisuru to omowa I invite that think nai, sunawati (watasi wa) John ga Mary 0 not that is I syootaisi nai to omou. invite not that think 'I don't think that John will invite Mary, that is, I think that John will not invite Mary.‘ 279 We observe that both (4.10) and (4.11) are grammatical. Next, let us negate one of (4.1) and (4.2), and <3onjoin it with the other as in: (4.12) *(watasi wa) John ga Mary 0 syootaisi nai to I invite not that omowa nai, sunawati (watasi wa) John ga Mary 0 think not that is I syootaisuru to omowa nai. invite that think not 'I don't think that John will not invite Mary, that is, I don't think that John will invite Mary.‘ (4.13) *(watasi wa) John ga Mary 0 syootaisuru to I invite that omou, sunawati (watasi wa) John ga Mary 0 think that iS I syootaisi nai to omou. invite not that think 'I think that John will invite Mary, that is, I think that John will not invite Mary.‘ 111 either case, the resulting sentence is unacceptable. Third, we observe that any of the following sentences IIlayoccur following both (4.1) and (4.2): (4.14) a. (watasi wa) Bill ga Mary 0 syootaisuru to I invite that omou think 'I think that Bill will invite Mary' 280 b. (watasi wa) John ga Jane 0 syootaisuru to I invite that omou think 'I think that John will invite Jane' c. (watasi wa) John ga Mary 0 musisuru to I ignore that omou think 'I think that John will ignore Mary' These observations will be sufficient to indicate the syno- nymity of sentences (4.1) and (4.2). Next, let us reconsider (4.3) and (4.4), repeated here as (4.15) and (4.16), respectively: (4.15)(=4.3) (watasi wa) John ga Mary 0 syootaisuru no I invite that de wa nai to omou. is not that think 'I think that it is not so that John will invite Mary.’ (4.16)(=4.4) (watasi wa) John ga Mary 0 syootaisuru to I invite that Ea omowa nai. think not 'I don't think that John will invite Mary.‘ We observe that sentences (4.15) and (4.16) are synonymous with sentences (4.1) and (4.2). This can be easily demonstrated in a way similar to the synonymity of (4.1) and (4.2). First, observe the grammaticality of the following 281 derived by conjoining (4.2) with (4.15) or (4.16): (4.17) (watasi wa) John ga Mary 0 syootaisuru to omowa I invite that think nai, sunawati (watasi wa) John ga Mary 0 not that is I syootaisuru no de wa nai to omou. invite that is not that think 'I don't think that John will invite Mary, that is, I think that it is not so that John will invite Mary.’ (4.18) (watasi wa) John ga Mary 0 syootaisuru to omowa I invite that think nai, sunawati (watasi wa) John ga Mary 0 not that iS I syootaisuru to wa omowa nai. invite that think not 'I don't think that John will invite Mary, that is, I don't think that John will invite Mary.’ Next, negating (4.2) and conjoining it with (4.15) or (4.16), we will get: (4.19) *(watasi wa) John ga Mary 0 syootaisuru to omou, I invite that think sunawati (watasi wa) John ga Mary 0 syootaisuru that is I invite no de wa nai to omou. that is not that think 'I think that John will invite Mary, that is, I think that it is not so that John will invite Mary.’ 282 (4.20) *(watasi wa) John ga Mary 0 syootaisuru to omou, I invite that think sunawati (watasi wa) John ga Mary 0 syootaisuru that is I invite to wa omowa nai. that think not 'I think that John will invite Mary, that is, I don't think that John will invite Mary.‘ Both (4.19) and (4.20) are unacceptable. Third, we observe that any of the sentences in (4.14) can occur following both (4.15) and (4.16). These observations indicate that (4.15) and (4.16) are synonymous with (4.2), and so with (4.1), because (4.1) and (4.2) are synonymous as noted above. If this line of argumentation is acceptable, we will have to account for the synonymity of (4.1), (4.2), (4.15) and (4.16) rather than the synonymity of (4.1) and (4.2) only. It is clear that the approach in terms of Negative—raising, applied to Japanese, may account for the synonymity of (4.1) and (4.2) but not the synonymity of (4.1), (4.2), (4.15) and (4.16). Our proposal here is that: (4.21) Sentences (4.15) and (4.16) are more closely related to (4.2) than to (4.1): in fact, (4.2), (4.15) and (4.16) are derived from the same underlying structure that is slightly different from the structure underlying (4.1). More Specifically, sentences (4.2), (4.15) and (4.16) will be derived from the underlying structure (4.22): 283 (4.22) /\VP\‘I” P NP'/ \7 N watisi ga omou IN: :2. pres I /// P\\\ AIX not N NP V pres I John ga N syootaisuru invite Mary 0 0n the other hand, sentence (4.1) will be derived from (4.23): (4.23) 8 MP ”/////’,VP‘\\\\\\\\‘ ATX N NP V pres I I watasi ga S2 omtu N NP Neg pres I I I I John ga N syootaisuru nai invite not Notice, in particular, that nai "not" in (4.22) is 284 sentential negation, while pg; "not" in (4.23) is verb— phrase negation. First of all, we must be able to account for the syno- nymity of sentence (4.1) with (4.2), (4.15) and (4.16). This is not difficult within our framework. Recall that we have proposed in (2.150) of Chapter II a presumably universal constraint that the scope of negation, whether sentential or verb-phrase, is the whole sentence in which it occurs. Thus, the scope of pa; in (4.22) is 82, while that in (4.23) is also 32: (4.24) [[[[John ga]NP [[Mary OJNP [syootaisuruIVJVP invite [preSIAuXIS JNP [EnaiJNngVP waJS 3 not 2 (4.25) [[John ga]NP [[Mary OJNP [syootaisuru]V [naiJNeglVP invite not [preSIAuXIS2 The comparison between (4.24) and (4.25) shows that the relevant parts included in the SCOpe of negation in (4.24) are the same as in (4.25). This explains why sentence (4.1) derived from (4.23) may be synonymous with sentences (4.2), (4.15) and (4.16) derived from (4.22). Furthermore, this explains why sentence (4.26) derived from (4.24) and sentence (4.27) derived from (4.25) are synonymous with each other: (4.26) [[John wa Mary 0 syootaisuru no]S de wa nai]S invite that is not 'It is not so that John invites Mary.‘ 285 (4.27) [John wa Mary 0 syootaisi pgils invite not 'John does not invite Mary.‘ Sentences (4.26) and (4.27) are synonymous in the following sense. First, conjoining (4.26) and (4.27) in terms of sunawati "that is," we will get a grammatical sentence, (4.28): (4.28) a. John wa Mary 0 syootaisuru no de wa nai, invite that is not sunawati John wa Mary 0 syootaisi nai. that is invite not 'It is not so that John invites Mary, that is, John does not invite Mary.‘ b. John wa Mary 0 syootaisi nai, sunawati John invite not that is wa Mary 0 syootaisuru no de wa nai. invite that is not 'John does not invite Mary, that is, it is not so that John invites Mary.' Second, conjoining (4.26) with an affirmative counterpart of (4.27) or conjoining (4.27) with an affirma— tive counterpart of (4.26), we will get an ungrammatical sentence in either case: (4.29) a. *John wa Mary 0 syootaisuru no de wa nai, invite that is not sunawati John wa Mary 0 syootaisuru. that is invite 'It is not so that John invites Mary, that is, John invites Mary.’ 286 b. *John wa Mary 0 syootaisi nai, sunawati John invite not that is wa Mary 0 syootaisuru no da. invite that is 'John does not invite Mary, that is, it is that John invites Mary.‘ Third, any of the following sentences can occur following both (4.26) and (4.27): (4.30) a. Bill ga Mary 0 syootaisuru invite 'Bill invites Mary' b. John wa Jane 0 syootaisuru invite 'John invites Jane' c. John wa Mary 0 musisuru ignore 'John ignores Mary' Thus, the difference between sentential and verb— phrase negation does not cause a semantic difference in cases such as (4.24) and (4.25). That is to say, in spite of some apparent difference between (4.24) and (4.25), sentences (4.26) and (4.27) derived from (4.24) and (4.25), respectively, may be synonymous because the scope of nega- tion is the whole sentence in which it occurs. If (4.26) and (4.27) derived from (4.24) and (4.25), respectively, are synonymous, it follows that sentences derived from (4.22) and (4.23), respectively, are synonymous. Thus, the synonymity of (4.1) with (4.2), (4.15) and (4.16) may be accounted for in terms of the putative universal constraint 287 regarding the SCOpe of negation, prOposed in (2.150) of Chapter II. Second, if sentences (4.2), (4.15) and (4.16) are derived from the same underlying structure, (4.22), their synonymity is automatically explained. Now, let us consider their derivation from the structure (4.22). If Sentence- raising does not apply to raise S3 into S2, the complementizer pp "that" and the copula dg ”is" are inserted into S and S , reSpectively, deriving sentence 3 2 (4.15). In turn, if S of (4.22) is raised into 82, it 3 gives the intermediate structure (4.31): (4.31) NP/ SmixifP\ I If/ \V I T pres watasi ga 82::::::\\‘\~ olou I VP wa think e- th John ga Mary 0 ‘ syootaisuru to nai invite that not Next, the negative nai is attached to the verb omou, giving: 288 NP//\\AIX I N NP/ pres watasi ga ///11Neg ohn ga Mary 0 omou syootaisuru to wa (4.32) The structure (4.32) becomes sentence (4.16). To cover the derivation of (4.32) from (4.31), we will need a rule such as: (4.33) x [w Negls Y vTh z—-> X [ W JS Y VTh+Neg Z where W, X, Y and Z are variables, Z contains no Neg, and Neg is exhaustively dominated by a VP Rule (4.33) is applicable to VTh, a small class of non- factive verbs of mental action such as pmgp "think, suppose," sinzuru "believe" and kangaeru ”think." This rule is very similar to the Negative-raising rule but the relevant dif- ference is that (4.33) is restricted to negatives which are exhaustively dominated by a verb phrase, that is, sentential negatives in our analysis. This is no small difference, as will be discussed in detail in the next section. To mention just one example, consider the following:3 289 (4.34) a. [(watasi wa) [kare ga sono hon sika yoma I he that book read nai to]S omou]S not that think 'I think that he will read only that book.’ b. *[(watasi wa) [kare ga sono hon sika yomu I he that book read tOJS omowa nai]S that think not 'I don't think that he will read _____ book.’ Negative-raising, applied to Japanese, will optionally apply to the structure underlying (4.34a), deriving (4.34b), which is not a grammatical sentence in Japanese. In contrast, rule (4.33) cannot apply to the structure underlying (4.34a), that is, (4.35): (4.35) "””fl””’p,,.»:;si-i\p“‘~‘\‘~““ NP VP Aux % Ya”””' ‘\\\\\\\‘TP p188 watasi ga omou I think ,,,//””/”Sg::;““‘-ilpp‘ NP 1////~\\\\\;_ kare ga sono hon sika yomu nai he that book read not // ___<;// It should be noticed that the negative in (4.35), which is not exhaustively dominated by a verb phrase, does not meet the condition of rule (4.33). Thus, the generation of 290 (4.34b) from (4.35) is blocked by the filtering function of this rule. Furthermore, considering (4.33), we note that it may be conjoined with Negative-attachment in (2.116). The conjoined rule will be: (4.36) a.(=2.ll6) { v } P Y Adj + wa Q + “a Y IAde V P Y {Adj}+wa+Neg X i V J z Q + wa Y IAdj + Neg b.(=4.33) X [ w Neg 18 Y VTh Z ——+ X [ W IS Y VTh+Neg Z where P, Q, N, X, Y and Z are variables, Z contains no Neg, P includes Ade or AdvD, Q includes Quant, Ade or AdvD, and Neg is exhaustively dominated by a VP It should be recalled that (4.36a) as well as (4.36b) is restricted to sentential negation, exhaustively dominated by a verb phrase. Furthermore, 'note the condition that Z contains no Neg. If Z contains a Neg and Negative— attachment (4.36b) applies, it derives an ungrammatical sentence as in the following: (4.37) [watasi ga [[John ga Mary 0 syootaisuru to I invite that JNP omou nai preSIS -—* [Enai] J wa] Neg VP 82 1 not think not 291 *[watasi wa [John ga Mary 0 syootaisuru to wa]S I invite that omowa nai nai]S think not not The same is true in the case of (4.36a). Turning back to sentence (4.16), the optional deletion of pg in (4.16) generates sentence (4.2), as desired. This pg-deletion may be compared with that relating sentences such as (4.38a) and (4.38b): (4.38) a.(=3.200a) kanozyo wa koohukuni pg sina nakatta. she happily die not-did 'She did not die happily.‘ b.(=3.177) kanozyo wa koohukuni sina nakatta. she happily die not-did 'She did not die happily.‘ It is to be noted that no new transformation is necessary, though we need some revision of Negative- attachment, to derive sentences (4.2), (4.15) and (4.16) from the same underlying structure, (4.22). This means that the analysis in terms of Negative-raising can be refined and incorporated into a more general analysis, that is, our analysis prOposed in Chapter II. Those transformations used in the derivation of (4.2), (4.15) and (4.16) are inde- pendently motivated to cover the derivation of sentences involving negation and quantifiers or certain classes of adverbials, as discussed in Chapter II and III. In this sense, our analysis is more general and well-motivated than the Negative-raising analysis. 292 4.2. Advantages pf the Prgposed Analysis 1 Japanese —— To confirm the point at the end of the preceding section, let us consider some clear advantages of our analysis. We note that there are a number of cases in which sentences of the type "NP wa [ X - Y 38 omowa nai/omotte inai (= NP do(es) not think [.X - Y IS)" are not synonymous with their counterparts of the type "NP wa [.X — Y - nai]S omou/omotteiru (= NP think(s) [ X - not - Y IS clear that Negative-raising, if it is to be a meaning- )." It is preserving transformation, cannot apply in these cases. A reasonable way to block its application in the cases will be to impose a constraint on the rule. Then, the approach in terms of Negative-raising will have to set up two different underlying structures for sentences of the two types in question, because they are not synonymous. Thus, the approach in terms of Negative-raising must distinguish two cases, cases where sentences of the two types are synonymous and those where they are not synonymous. Moreover, if sentences of one type are grammatical and their counterparts of another type are ungrammatical, they pose another problem of a similar kind for Negative-raising. Clearly it will be preferable if we can treat them in a unitary way. In this section, we will cite a number of cases in which sentences of the two types are different in meaning or grammaticality and demonstrate that our analysis can explain them in the 293 same way as those cases in which sentences of the two types are synonymous. Some characteristic cases where sentences of the two types are not synonymous involve negation and quantifiers or those classes of adverbials cited in Chapter III. First, let us observe sentences such as (4.39) and (4.40) involving a quantifier: (4.39) (watasi wa) subeteno oobosya ga sintaikensa o I all applicant physical- examination uke nakatta to omou. undergo not-did that think 'I think that all the applicants did not undergo a physical—examination.‘ (4.40) (watasi wa) subeteno oobosya ga sintaikensa o I all applicant physical- examination uketa to omowa nai. underwent that think not 'I don't think that all the applicants underwent a physical-examination.‘ Sentence (4.39) is clearly not synonymous with (4.40). This is due to the fact that the negation in (4.39) does not include subeteno "all" in its scope, while that in (4.40) includes subeteno "all" in its s00pe. Under our analysis, if the negation in (4.39) is verb-phrase negation, that in (4.40) is sentential negation. In this connection, further compare the following examples with (4.39) and (4.40): 294 (4.41) (watasi wa) subeteno oobosya ga sintaikensa o I all applicant physical- examination uketa no de wa nai to omou. underwent that is not that think 'I think that it is not so that all the applicants underwent a physical-examination.‘ (4.42) (watasi wa) subeteno oobosya ga sintaikensa o I all applicant physical- examination uketa to Hg omowa nai. underwent that think not 'I don't think that all the applicants underwent a physical-examination.‘ We observe that (4.41) and (4.42) are synonymous and they are in turn synonymous with (4.40). Thus, our proposal is that (4.40), (4.41) and (4.42) are derived from one underlying structure, and (4.39) is derived from another. Applying the analysis of the preceding section, we hold that sentences (4.40), (4.41) and (4.42) are derived from the underlying structure (4.43), whereas sentence (4.39) is derived from (4.44): (4. 431:?// 1\Aux | NP/ H‘IIIIIIIIII“ WI watasi J omou I 11V//’2""‘-=::::::::::::::~~_\~ think ////S3::::::::::73II‘II“‘Aux :Ef ///yE\\\\ /// \\\\\V I not Quant N NP past subeteno oobosya N ukeru all applicant I undergo sintaikensa 0 ga physical-examination (4.44)W/81\Alilx I / \ N V pres watasi omou I llx//ISM.IIIIIII‘III“‘*--.1._‘Pthink ga Quant SIB.\-IIIIIIIIIm/‘VP-INNI'IN“Aux\ subeteno ////‘\\\ /¥:\\\\WN all Quant N NP eg past I I I A oobosya N ukeru nai applicant I undergo not sintaikensa 0 ga physical-examination 296 The structure (4.43) indicates that the negative nai, whose scope is 32, includes the quantifier subeteno "all" in its scope. In contrast, the structure (4.44) shows that the 3 2 is outside of its SCOpe. This difference reflects the semantic scope of the negative is S and the quantifier in S difference between the sentences derived from them. Now, considering the derivation involving (4.43), if Sentence- raising does not apply, the complementizer no "that," and the copula da "is" and the complementizer to "that" are inserted into S3 and 82, respectively, yielding sentence (4.41). In turn, if S is raised up into 82, it derives the 3 intermediate structure: \ T13 /VP\ Aux N NP V pies watlsi ga 42 olou ’,r////”’ ‘:E$EEEEEEEF::::::-‘-‘-‘ NP VP Aux VP wa /\N /\ I I Quant NP V past Neg l I I 1 subeteno oobosya N ukeru nai ga sintaikensa 0 Then, applying Negative-attachment (4.36b) to attach nai ”not" to omou "think” as well as the complementizer- insertion rule, we derive: 297 (4.46) S NP"””’//””””fl i::::;;:::::::::::j‘-~Aux N NP’/’I/z’ff” V pies watisi ga A2 \\\Eeg NP/ my: I \\\\ //’ ‘\\ a omou nai Quant N NP 1 subeteno oobosya N uketa ga underwent sintaikensa o The structure (4.46) generates sentence (4.42). Then, the optional deletion of HE gives sentence (4.40). On the other hand, if the quantifier in (4.45) is optionally shifted to follow the subject noun phrase and the contrastive Ea is attached to the quantifier in a way discussed in section 2.6, the following intermediate structure is derived: (4.47) /Sl\ :13 NP/\0P\ ATX V pres l 1 b wataSi g8l/ 2\\\m NP VP Aux VP /\ /\l N Quant TP V past Neg oobosya subete a N uklru nai g8. sintaikensa o 298 Then, Negative-attachment (4.36a) can apply to attach nai "not" to the verb ukeru "undergo," deriving: NP / P\ Alllx N /T:\ T pres /P ATX NP /\\\\\V past oobosya ga subete £ \\\ Neg l sintaikensa o ukeru nai The structure (4.48) yields sentence (4.49): (4.49) (watasi wa) oobosya ga subete Ea sintaikensa o I applicant all physical— examination uke nakatta to omou. undergo not-did that think 'I think that not all the applicants underwent a physical-examination.’ Next, let us turn to the derivation of (4.39) from (4.44). First, S3 is raised into 82 and the quantifier subeteno "all" is attached to A in a way discussed in section 2.9.1, deriving the intermediate structure (4.50): 299 (4.50) / sl 1\ NP Aux l L HUI////”//’ ‘\\\\\\\\\‘V pres watasi ga omou NP/I 2\ think I///, /////V\ Aux Quant \\\~T TP %\ Neg pist sublteno oobosya ga N ukeru nai all applicant l undergo not sintaikensa o physical-examination The structure (4.50) becomes sentence (4.39). It should be noted that the negative in (4.50) has no chance of combining with the verb 9mgu "think" to derive sentence (4.40), since it is not exhaustively dominated by a verb phrase. As stated in Negative—att achment in (4. 36), negatives must be exhaus- tively dominated by a verb phrase, that is, they must be sentential negatives to combine with a higher verb of mental action such as gmqu. Furthermore, notice that the condition of Sentence-raising is satisfied in (4.50), since the higher verb gmgu "think" belongs to the class of verbs of saying and thinking. Thus, the Optional application of Sentence-raising gives the intermediate structure: (4. 51) fl 51\\ In NP VP VP Alllx N Quant \\\\\N’ 2/////\\\\\\\ V' pres l sintaikensa o l watasi subeteno oobosya omou uke nakatta not-did ga ga Then, attaching the adverbial marker g; to omou "think" and moving it to follow the subject watasi gg in a way discussed in section 2.6, we get sentence (4.52): (4.52) watasi gg omou gi, subeteno oobosya ga I think all applicant sintaikensa o uke nakatta. physical-examination undergo not-did 'All the applicants, I think, did not undergo a physical—examination.’ From the foregoing discussion we see that there is no chance of generating sentence (4.40) from the structure underlying (4.59). Thus, our analysis can explain why (4.59) is not synonymous with (4.40) as well as account for the synonymity of sentences (4.40), (4.41) and (4.42) without introducing any new transformation. Furthermore, we can relate sentences such as (4.49) and (4.52) to (4.40) and (4.39), reSpectively. The second major type involves negation and the adverbials discussed in Chapter III, specifically, nominal adverbials, adverbials of frequency and adverbials of reason and purpose. To take a concrete example, consider sentences such as (4.53) and (4.54) involving a nominal adverbial: 501 (4.55) (watasi wa) sono syoonen dake ga ohiru o tabe I that boy only lunch eat nakatta to omou. not-did that think 'I think that only that boy did not eat lunch.‘ (4.54) (watasi wa) sono syoonen dake ga ohiru o tabeta I that boy only lunch ate to omowa nai. that think not 'I don't think that only that boy ate lunch.’ Clearly (4.55) is not synonymous with (4.54). Moreover, compare them with (4.55): (4.55) a. (watasi wa) sono syoonen dake ga ohiru o I that boy only lunch tabeta no de wa nai to omou. ate that is not that think 'I think that it is not so that only that boy ate lunch.’ b. (watasi wa) sono syoonen dake ga ohiru o I that boy only lunch tabeta to 13 omowa nai. ate that think not 'I don't think that only that boy ate lunch.‘ We observe that (4.55a) and (4.55b) are synonymous and they are in turn synonymous with (4.54), not (4.55). Under our analysis, the structure underlying (4.54) and (4.55) is (4.56), while that underlying (4.55) is (4.57): 502 (4.56) M/SlvAux & W”/’,,/’V P~\\\\\\\\‘ V pres watasi ga I ma \\ think NP/\\pm 1:: //‘/\N /\ Ava D t NP \V past not dike sono syoLnen N taberu only that boy ohiru 0 eat ga lunch (4.57) s m/ 1v” N N””’,’V P‘\\\\\\\‘Y‘ pres l l watasi ga —ffly—fly’”,._~..—--32....“\“\\\‘~ omou I :P VP think AJV ,/ S5t::::::::::f-~___‘______ I N NP Aux dake ’,,/z;, \\\\ //V \\~ I only Ava Dit N NP V eg past I A sono syoonen N taberu nai that boy ohiru 0 eat not ga lunch The structure (4.56) shows that the negative Egg, whose SCOpe is 82, includes the adverbial dgkg "only" in its scope, whereas the structure (4.57) indicates that the scope of the negative is S3 and the adverbial ngg is outside of its sc0pe. This relevant difference accounts for the semantic 303 difference between the sentences derived from them. Now, considering the derivation involving (4.56), the insertion of the complementizer pg "that" into 83’ and the copula da "is" and the complementizer to "that" into 82 generates sentence (4.55a). If, on the other hand, Sentence-raising applies to raise 83 up into 82, it derives the following intermediate structure: (4.58) -fl'—I.—I”IflI—.-m.fl____________.....——-—-'Sl-2::u:::::‘~__mm_“~~“~ NP VP Aux l \ l N NP V pres watasi ga ',,,”””’gflflS§~\“““~‘l-:%fu NP \VI\VP 4///’\\\\\\i 4//////\\\\\\;‘ NLg sono syoonefl ohiru o tabeta I dake ga ate nai Then, applying Negative-attachment (4.36b) as well as the complementizer-insertion rule, we get (4.59): (4.59) .. //Sl\\‘ NP VP Aux bl: NP/ \v I ' W l watasi ga ””’,z,"32 \\ NP VP\\\to wa omou nai sono syoonefl’ ohiru o dake ga tabeta The structure (4.59) becomes sentence (4.55b). Then, the Optional deletion of ya gives sentence (4.54). 304 Turning next to the derivation of (4.53) from (4.57), Sentence-raising applies to raise S3 into 82 and Quantifier- attachment assigns dake "only" to A, giving the intermediate structure (4.60): NP P Aux I NP \ pres watasi ga 42 omou Nf:::::::””’fl Ti:E:VP::::f-“Aux Adv Det N NP t Neg past I“! I l I dake sono syoonen N taberu nai ga ohiru o Inserting the complementizer :2 "that" into 82, we derive sentence (4.53). It is to be noted that since Neg in (4.60) is not exhaustively dominated by a verb phrase, Negative- attachment cannot apply to attach the negative to the verb omou "think." Thus, the derivation of sentence (4.54) from (4.60) is correctly blocked through the filtering function of Negative-attachment. We further note that the condition of Sentence-raising is satisfied in (4.60). Applying Optionally this transformation, after the application of the relevant transformations in the S2-cycle, we derive (4.61): 305 (4.61) _fifl’_,,,.———e~*“""Si““‘--—~___~‘_‘_‘_ NP ’////’ :::;5;-‘~‘~“"VP Aux I l////\\\\\\‘ I I V pres sono syoonen 'Ehiru o tabe I watasi ga dake ga nakatta omou not-did The structure (4.61) generates sentence (4.62), which is synonymous with (4.53): (4.62) watasi ga omou _i, sono syoonen dake ga ohiru I think that boy only lunch o tabe nakatta. eat not-did 'Only that boy, I think, did not eat lunch.’ The foregoing discussion demonstrates that our analy- sis can explain why sentences (4.53) and (4.54) are not synonymous as well as account for the synonymity of (4.54) with (4.55). A similar argument holds for adverbials of frequency and adverbials of reason and purpose with respect to nega— tion. To mention one example for each, consider next the following involving an adverbial of frequency: (4.63) (watasi wa) Bill ga itumo yakusoku o mamora I always promise keep nakatta to omou. not-did that think 'Lit. I think that Bill always did not keep his promise.‘ 306 (4.64) (watasi wa) Bill ga itumo yakusoku o mamotta I always promise kept to omowa nai. that think not 'I don't think that Bill always kept his promise.’ Clearly (4.63) and (4.64) are not synonymous. Next, compare them with (4.65): (4.65) a. (watasi wa) Bill ga itumo yakusoku o mamotta I always promise kept wake de wa nai to omou. that is not that think 'I think that it is not so that Bill always kept his promise.’ b. (watasi wa) Bill ga itumo yakusoku o mamotta I always promise kept to ya omowa nai. that think not 'I don't think that Bill always kept his promise.‘ We observe that (4.65a) is synonymous with (4.65b) and that they are synonymous with (4.64), not (4.63). Within our framework sentences (4.64) and (4.65) are derived from the underlying structure (4.66), whereas (4.63) is derived from (4.67): 307 NT I pres watasi ga ",,»”””Si““\~\\\\\‘ omou I In) \V‘P W8. think S Neg NP WP nai \ NP ‘//VP\\ Aux itumo l N NP V past always Bill ga N mamoru yakusoku 0 keep promise NP VP Aux l watasi ga 82 omou I NP/ \ think ’3 l 1 Bill ga N mamoru nai VP l l S AdvF NPr”///" 3:VP\~\\\\\‘Aux itumo I / I N V \Niag paIst always yakusoku 0 keep not promise Observe the relevant difference between them: in (4.66) the negative, whose SCOpe is 82, includes the adverbial itumo 308 "always" in its scope, whereas in (4.67) the scope of the negative is 83 and the adverbial is outside of its scope. Now, let us turn to the derivation involving (4.66). 84 is first raised into 83 and the adverbial is moved to follow the subject noun phrase. Then, the insertion of the complementizer wake "that" into 83’ and the 00pula Q; "is" and the complementizer to "that" into 82 gives sentence (4.65a). If, on the other hand, Sentence-raising further applies to raise S3 into 82, it derives (4.68): (4.68) ‘”,,,,,’r*"’”"flfl'SI::::::T“--~._____- NP My/(fyflvVP‘.\\\\\\~ Aux N NP V pres watasi gi”’#i:::;::;fi’?2 omou NP AdvF VP :::::AEE7“VP I . I (5’ \\\ I I N itumo I Y past Neg BiIl ga N mamoru nai yakusoku 0 Next, Negative-attachment (4.36b) applies as well as the insertion of the complementizer to "that” into S (4.69): 2, giving 309 ( .6 ) 4 9 NP’fl’flflflflfla'flflflfl'fl#fli:iEfEE;;g:~“~‘§‘-‘““Aux N :P ‘\\\\\\\‘V pies watasi ga / Neg IF AfivF /\§P‘\‘to\\\‘a omou nLi N itmm Ng/P\\Wo 1 Bill ga N mamotta yakusoku o kept The structure (4.69) becomes sentence (4.65b). Then, the optional deletion of ya gives sentence (4.64). Turning next to the derivation of (4.63) from (4.67), Sentence-raising applies to raise 83 into 82, giving the intermediate structure (4.70): ( . ) 4 70 NP/ SlYAM N l”///” ‘\\\\\‘V I pres IIP \Aux N I ! Neg past Ang BilI ga N mamoru nai itumo yakusoku 0 Then, the adverbial is shifted to follow the subject noun phrase, giving sentence (4.63). We note that since Neg in (4.70) is not exhaustively dominated by a verb phrase, Negative-attachment (4.36b) cannot apply to attach the 310 negative to the verb gggu "think." Thus, the generation of sentence (4.64) from (4.70) is blocked through the filtering function of Negative-attachment. We further note that the structure (4.70) meets the condition of Sentence-raising, because the verb QQQE belongs to the class of verbs of saying and thinking. The optional application of Sentence-raising, after the shift of the adverbial to follow the subject noun phrase, gives (4.71): (4.71) s M I\ / \VP ~ NP NP Adv P ux I, 1 1F 1 N N itumo V pres L yakusOku o mamora I wa asi ga Bill ga nakatta omou not-did The structure (4.71) generates sentence (4.72), which is synonymous with (4.63): (4.72) watasi ga omou i, Bill wa itumo yakusoku o I think always promise mamora nakatta. keep not-did 'Bill, I think, always did not keep his promise.‘ Next, consider sentences such as (4.73) and (4.74) involving an adverbial of reason: (4.73) (watasi wa) kare ga sore gg riyuu‘dg kekkonsi I he that reason for get married nakatta to omou. not-did that think 'Lit. I think that for that reason he did not get married.‘ 311 (4.74) a. (watasi wa) kare ga sore ga riyuu dg I he that reason for kekkonsita to omowa nai. got married that think not 'Lit. I don't think that he got married for that reason.‘ b. (watasi wa) kare ga sore ga riyuu dg I he that reason for kekkonsita no de wa nai to omou. got married that is not that think 'Lit. I think that it is not so that he got married for that reason.‘ c. (watasi wa) kare ga sore ga riyuu.d§ I he that reason for kekkonsita to wa omowa nai. got married that think not 'Lit. I don't think that he got married for that reason.’ Comparing them, we see that the sentences of (4.74) are synonymous with each other, but not with (4.73). Under our analysis, the sentences in (4.74) have the same underlying structure, which is different from that underlying (4.73). More specifically, sentences (4.74) and (4.73) are derived from (4.75) and (4.76), reSpectively: (4.75) 3 ~ Nfi"'dflflrffffl’flflfl‘ i::::::E;““I“‘-~AuX I NP/ \V I N pres t I S' or'nou W8. .38]. 83 / \ I NP 2\\\“\VF wa think /;3\ Nig- NP VP n i ' Adv not S 4 NP/ filxu /\ It (I ‘ past sore ga riyuu de l l kare ga kekkonsuru that reason for he get married (4.76) ”,,,,’«"””flflSi7fi““‘-~al_““““~‘ V ~:::::::VP Aux I NP \Vl, I N I pres wataSi ga 82 \ omou I NP VP think NP VP Aux I //// \\\ l .i N Y Nfg past sore ga riyuu de kare ga kekkonsuru nai that reason for he get married not The structure (4.75) indicates that the negative, whose scope is 82, includes the adverbial in its s00pe, whereas the structure (4.76) shows that since the scope of the negative is 83' the adverbial is outside of its scope. This difference reflects the semantic difference between (4.74) 313 and (4.73). Now, let us consider the derivation involving (4.75). After the raising of S4 into S3 and the shift of the adver- bial to follow the subject noun phrase, the insertion of pg "that" into 83' and da "is" and EQ "that" into 82 gives sentence (4.74b). If, on the other hand, Sentence—raising further applies to raise S into S the intermediate 3 2’ structure (4.77) is derived: (4.77) //Sv NP VP Aux ) / N NP \\\\‘Y pies watasi.giflflfiflfl_fl_____::::::::rS§~“““§§‘~:fff:~ NP Adv W VP wa I R I N . Nfg kare ga sore ga riyuu de 'EEkkonsita nai got married Next, applying Negative-attachment (4.36b) and the complementizer-insertion rule, we get (4.78): (4 78) m_.,——"””'flfflfififlfiis f\“--‘;~‘““r-~___‘_m N W””/””’—V P‘\\\\\\\\\"V pres l l Vp\ watasi ga 82 Neg l __l.ea—*""flflflfl‘fl’,~‘---~““--l_w omou nai kare ga sore ga riyuu de kekkonsita to wa The structure (4.78) generates sentence (4.740). Then, the Optional deletion of ya gives sentence (4.74a). 314 Turning next to (4.76), the application of Sentence- raising and Adverbial-movement to shift the adverbial to follow the subject noun phrase derives the intermediate structure: NP \VP\ Alix N NP Y pres watisi ga "’f"”””,,,.as omou 2\ NP AdvR/ VP 1 N /\ /\ l ‘ kekkonsi nakatta kare ga sore ga riyuu de not-did The structure (4.79) yields sentence (4.73). Moreover, Sentence-raising can further apply to raise S2 into 81’ giving: ”'80)/////"S€§\\ NP l Adv VP VP Aux IP I | 4///////Aéi\\\_ J ‘ N N pres I I kekkonsi \ watasi ga kare ga sore ga riyuu He nakatta omou Then, the structure (4.80) generates sentence (4.81), which is synonymous with (4.73): (4.81) watasi ga omou pi, kare wa sore ga riyuu de I think he that reason for kekkonsi nakatta. get married not-did 'For that reason, I think, he did not get married.’ 315 The foregoing discussion shows that no new relevant transformation is necessary to take care of the cases in which sentences of the two types in question are not synony- mous when they involve quantifiers or certain classes of adverbials. 1 We further note that there are a number of cases in which sentences of one type are grammatical, but their counterparts of another type are not. Such a discrepancy in grammaticality between sentences of the two types is observed, for instance, when they involve the Japanese equivalents of the English incomplete negatives such as seldom, hardly, rarely, only, few and little. As already discussed in section 2.1, the Japanese counterparts of the English incomplete negatives consist of "adverbial + negative" and this adverbial cannot occur in affirmative environments. To illustrate with a concrete example, observe the following: (4.82) a. Tom wa furansugo sika hanasa nai. French speak not 'Tom speaks only French.’ b. *Tom wa furansugo sika hanasu. French speak In (4.82a) "sika + nai" as a unit corresponds to gply, so there is no equivalent of the adverbial sika in English. In (4.82b) sika occurs in an affirmative environment, which makes the sentence ungrammatical. Now, consider the following: 316 (4.83) [(watasi wa) [Tom ga furansugo sika hanasa nai I French Speak not tOJS omou]S that think 'I think that Tom speaks only French.’ (4.84) *[(watasi wa) [Tom ga furansugo sika hanasu to]S I French Speak that omowa nai]S think not 'Lit. I don't think that Tom speaks French.’ Sentence (4.84) sounds ungrammatical, as is expected from the ungrammaticality of (4.82b). If we assume, following R. Lakoff (l969a), Lindholm (1969) and others, that (4.84) is derived from the structure underlying (4.83) by the appli- cation of Negative—raising, we cannot account for the difference in grammaticality between (4.83) and (4.84). According to our analysis, on the other hand, sentences (4.83) and (4.84) are derived from different structures such as (4.85) and (4.86), respectively: (4.85) s NP/// wilux % jT//’/’/’ Y pies watasi ga omou I NP/ 2\VP\‘Aux think NP Neg pres / L7 OD _- l N Tom ga hanisu nai I furaisugo sika speak not French Er IiM/VP.\ lulu: W8. a: i a O mou : g 1? v””””’ Sz\\\““;-~wa think m”////,’S 3>\\;\\\\\\\‘A hjfg 1‘ /P\\\\ Al nai N I? V pres not Tom gaN hanasu speak furansugo Sika French It should be observed that in (4.86) the adverbial sika occurs in S3 containing no negative, which causes the ungrammaticality of the resulting sentence (4.84). In contrast, in (4.85) gika occurs in S2 containing a negative, thus deriving the grammatical sentence in (4.83). Then, to block the generation of ungrammatical sentences such as (4.84) from (4.86), we need a constraint as follows: (4.87) The Japanese incomplete negatives such as mettani, hotondo and sika cannot occur in a simplex affirmative sentence. This constraint is in any case necessary to block the gener— ation of sentences such as (4.82b). 'Moreover, the constraint also blocks the generation of the following ungrammatical sentences from (4.86): 318 (4.88) a. *[(watasi wa) [[Tom ga furansugo sika hanasu I French speak nOJS de wa nai to]S omou]S that is not that think 'Lit. I think that it is not so that Tom speaks French.‘ b. *[(watasi wa) [Tom ga furansugo sika hanasu I French speak to]S wa omowa nai]S that think not 'Lit. I don't think that Tom speaks French.‘ Thus, our analysis can take care of the difference in gram- maticality between (4.83) on the one hand and (4.84) and (4.88) on the other. It is clear that the approach in terms of Negative—raising will need to impose another constraint on the rule to block its application to structures underlying sentences like (4.83) to derive (4.84). A few more similar examples may be cited in the following:4 (4.89) a. [sensei wa [Mary ga mettani naka nai to]S teacher weep not that omotteiru]S think 'The teacher thinks that Mary seldom weeps.’ b. *[sensei wa [Mary ga mettani naku to]S teacher weep that omotte inai]S think not 'Lit. The teacher doesn't think that Mary weeps.’ 319 (4.90) a. [(watasi wa) [musuko ga hotondo benkyoosi I son study nai to]S omou]S not that think 'I think that my son studies little.‘ b. *[(watasi wa) [musuko ga hotondo benkyoosuru I son study to]S omowa nails5 that think not 'Lit. I don't think that my son studies ____.' Constraint (4.87) can block the generation of ungrammatical sentences such as (4.89b) and (4.90b), and also (4.91) and (4.92) which will otherwise derive from the structures underlying (4.89b) and (4.90b), respectively: (4.91) a. *[sensei wa [[Mary ga mettani naku nOJS de wa teacher weep that is nai to]S omotteiru]S not that think 'Lit. The teacher thinks that it is not so that Mary weeps.‘ b. *[sensei wa [Mary ga mettani naku to]S E2 teacher weep that omotte inai]S think not 'Lit. The teacher doesn't think that Mary weeps.’ 320 (4.92) a. *[(watasi wa) [[musuko ga hotondo benkyoosuru I son study no]S de wa nai to]S omou]S that is not that think 'Lit. I think that it is not so that my son studies ____,’ b. *[(watasi wa) [musuko ga hotondo benkyoosuru I son study to]S Ea omowa nai]S that think not 'Lit. I don't think that my son studies .' In addition to those adverbials, there are some other words in Japanese which cannot occur in affirmative environ- ments, such as daremo ”anybody," nanimo "anything,"‘kessite "ever," and tootei "(not) possibly.” They may be included in the category of incomplete negatives here. A similar difference in grammaticality is observed in the cases involving these words, for instance, between (4.93) and (4.94): (4.93) [(watasi wa) [daremo sono kawa o koe nakatta I anybody the river cross not—did to]S omou]S that think 'Lit. I think that anybody did not cross the river. (= I think that nobody crossed the river.)' (4.94) *[(watasi wa) daremo sono kawa o koeta to]S I anybody the river cross that omowa nai]S think not 321 'Lit. I don't think that anybody crossed the river.’ Within our framework, sentences (4.93) and (4.94) are derived from (4.95) and (4.96), respectively: \v NP P Aux N NP Y pres watasi ga S NP VP Aux I ,,/”’ \:\\\\ | N NF V Neg past l // ‘\\ i I. daremo Det N koeru nai anybody cross not sono kawa o the river (4.96) NP//S \Al‘m N /\\\\\“ I IT watasi ga 32 omou NP VP wa think N V pres NP/ Aux nai l NP/V N P\\\\\V past not I I” m\\\ l daremo Det N koeru anybody ‘ I cross sono kawa o the river 'The structure (4.96) indicates that daremo occurs in a sentenCe containing no negative. Thus, constraint (4.87) 322 excludes it as an ill-formed structure, thereby blocking the generation of ungrammatical sentences such as (4.94) and also (4.97) which will otherwise derive from (4.96): (4.97) a. *[(watasi wa) [[daremo sono kawa o koeta no]S I anybody the river crossed that de wa nai to]S omou]S is not that think 'Lit. I think that it is not so that anybody crossed the river.‘ b. *[(watasi wa) [daremo sono kawa o koeta tOJS I anybody the river crossed that Ea omowa nails think not 'Lit. I don't think that anybody crossed the river.‘ A similar discrepancy in grammaticality is observed between the paired sentences in (4.98) through (4.100): (4.98) a. [anata wa [nanimo okora nakatta to]S you anything happen not-did that omotteiru]S think 'Lit. You think that anything did not hap en. (= You think that nothing happened. ' b. *[anata wa [nanimo okotta to]S omotte'inai]S you anything happened that think not 'Lit. You don't think that anything happened.’ 323 (4.99) a. [(watasi wa) [Jones ga kessite wareware o I ever us uragira nai to]S omou]S betray not that think 'Lit. I think that Jones will not ever betray us. (= I think that Jones will never betray us.)' b. *[(watasi wa) [Jones ga kessite wareware o I ever us uragiru to]S omowa nai]S betray that think not 'Lit. I don't think that Jones will ever betray us.’ (4.100) a. [kare wa [kimino risoo ga tootei zitugensare he your ideal possibly is realized nai to]S omotteiru]S not that think 'Lit. He thinks that your ideal will not possibly be realized.‘ b. *[kare wa [kimino risoo ga tootei zitugensareru he your ideal possibly is realized to]S omotte inai]S that think not 'Lit. He does not think that your ideal will possibly be realized.‘ Constraint (4.87) can also block the generation of ungram- matical sentences such as (4.101), (4.102) and (4.103) which will otherwise derive from the structures underlying (4.98b), (4.99b) and (4.100b), respectively: 324 (4.101) a. *[anata wa [[nanimo okotta no]S de wa nai you anything happened that is not to]S omotteiru]S that think 'Lit. You think that it is not so that anything happened.‘ b. *[anata wa [nanimo okotta to]S ya omotte you anything happened that think inai]S not 'Lit. You don't think that anything happened.‘ (4.102) a. *[(watasi wa) [[Jones ga kessite wareware o I ever us uragiru no]S de wa nai to]S omou]S betray that is not that think 'Lit. I think that it is not so that Jones will ever betray us.‘ b. *[(watasi wa) [Jones ga kessite wareware o I ever us uragiru to]S ya omowa nai]S betray that think not 'Lit. I don't think that Jones will ever betray us.‘ (4.103) a. *[kare wa [[kimino risoo ga tootei he your ideal ‘ possibly zitugensareru no]S de wa nai to]S omotteiru]S is realized that is not that think 'Lit. He thinks that it is not so that your ideal will possibly be realized.’ 325 b. *[kare wa [kimino risoo ga tootei he your ideal possibly zitugensareru to]S Ea omotte inai]S is realized that think not 'Lit. He doesn't think that your ideal will possibly be realized.‘ We may also add mgdg "until" here in that it cannot occur in affirmative environments. For example, observe the ungrammaticality of (4.105), as Opposed to the grammaticality of (4.104): (4.104) otooto wa kuzi made oki nakatta. brother nine until get up not-did 'My brother did not get up until nine.‘ (4.105) *otooto wa kuzi made okita. brother nine until got up 'Lit. My brother got up until nine.‘ Thus, a similar difference in grammaticality is observed between (4.106) and (4.107): (4.106) [(watasi wa) [otooto ga kuzi made oki I brother nine until get up nakatta to]S omou]S not-did that think 'I think that my brother did not get up until nine.’ (4.107) *[(watasi wa) [otooto ga kuzi made okita to] I brother nine until got up that S . 6 omowa nai]S think not 'Lit. I don't think that my brother got up until nine.‘ 326 According to our analysis, (4.106) and (4.107) are derived from different underlying structures such as (4.108) and (4.109), respectively: (4.108) —’”’flflflflflflfiflflg,,,l.—sl::::::~‘-~‘~“~“§~ NP VP Aux l M,z/”” ‘\\‘\\\‘ I N V pres l watasi ga /Ad/”’/’S“\\N\\\“~\A omou I IX think AdvT V//YP ///~\\\\\ I Nfg past otooto ga kuzi made okiru nai brother nine until get not up (4.109) _’_'~,,,,...———"'“'Si NP :::éifEEE;;;:::::::“t-Aux N NP v I l pres l watasi ga I”//’/,,,//—S§~\\\\\\N\\\~ omou I \\\\ wa think I Nfg NP fldv;\~“~‘~‘VF‘~‘Am nai l N Z///\\\\\x past not I I otooto ga kuzi made okiru brother nine until get up {Fhe structure (4.109) indicates that made "until" occurs in a sentence containing no negative, thereby violating the cxnuatraint in (4.87). Thus, constraint (4.87) excludes (4,]jx9) as an ill-formed structure, and by so doing it 327 blocks the generation of ungrammatical sentences such as (4.107) and also (4.110) which will otherwise derive from (4.109): (4.110) a. *[(watasi wa) [[otooto ga kuzi made okita I brother nine until got up no]S de wa nai to]S omou]S that is not that think 'Lit. I think that it is not so that my brother got up until nine.‘ b. *[(watasi wa) [otooto ga kuzi made okita I brother nine until got up to]S Ea omowa nai]S that think not 'Lit. I don't think that my brother got up until nine.’ Thus, our analysis based on the analysis prOposed in Chapter II can account for the difference in grammaticality between sentences of the type "NF wa [ X - Y - nai JS omou/omotteiru (= NP think(s) [ X - not - Y JS)" and the type "NP wa [ X - Y 38 omowa nai/omotte inai (= NP do(es) not think [ X - Y 18)" as well as account for the semantic difference between sentences of the two types involving negation and quantifiers or those adverbials cited in Chapter III. What is more, our analysis can treat them in the same way as those cases in which the two types are synonymous: one type involves verb-phrase negation, whereas the other type involves sentential negation, regardless of whether they are synonymous or not. In addition, our analysis can take care of them without resorting to any new 328 transformation, that is, in a relatively general way. In contrast, if we are to account for them in terms of the Negative—raising rule, a very complicated constraint must be imposed on the rule to block its application in these cases. Moreover, the approach in terms of Negative- raising will have to treat them in a different way from those cases where the two types are synonymous: only when sentences of the two types are synonymous may they be derived from the same underlying structures by the appli- cation of Negative-raising. That is to say, in the above-discussed cases where sentences of the two types are not synonymous or are different in grammaticality, they will have to be derived from different underlying structures without using the rule of Negative-raising. Thus, this approach has a number of disadvantages, or put differently, our analysis has a number of advantages over the approach in terms of Negative-raising. 4.3. The Application g: the PrOposed Analysis 39 English with Respect 39 Negative- Raising In this section we will apply the analysis of the preceding sections to English sentences of the type "NP think(s) [ X - not - Y 33" and the type "NP do(es) not think [X - Y IS." To start with, consider the following: (4.111) I think that he will ggt read a paper there. (4.112) I don't think that he will read a paper there.7 329 If we accept Negative-raising as a syntactic rule, they will be derived from the same underlying structure: ( . ) 4 llfmA//./81\VP I Y1 V/ \ WP N pres I I I . \ not N will Y there I 4: 3:: he read a pape The optional raising of the negative into Sl yields sentence (4.112). Otherwise, it gives sentence (4.111). Thus, this approach can account for the synonymity of sentences (4.111) and (4.112) in a natural way. But there are a number of cases in which this analysis is inapplicable, as will be discussed in the next section. On the other hand, applying our analysis in the preceding sections, we derive sentences (4.111) and (4.112) from different underlying structures such as (4.114) and (4.115), respectively: 330 NP A ”///’ \\yp ux N pres Y NP I think S NP Aux VP Adv l / \ L N will Neg Y\\\\\‘NP there I he not read a paper pres V NP think S 11W V?” s? Nfg NP Aux P Adv not I I \\\NP IL N will Y there I he read a paper Thta derivation of (4.111) from (4.114) is too straightfor- wazxi to need any explanation. As for the derivation of (4.Ill2) from (4.115), the application of Sentence-raising to raijse S3 into 82 gives the intermediate structure: 331 (4.116) s NP Aux VP N pres Y NP I in S NP Aux r/;YP Adv; VP I \n 1 I I N will V there Neg h e read a pape not Then, Negative-attachment applies to attach the negative to the verb think, deriving: (4.117) A//'Sl\vp ‘1" v / \. NP I 3 NBA I .// 2% pres HF‘2——'% not think NP Aux Adv I / \ L N will V NP there I he read a paper This becomes sentence (4.l12). To take care of the deriva- tion of (4.117) from (4.116), we need a rule such as: (4.118) X VTh [ W Neg 38 Y ——9 X Neg+VTh [ W 38 Y where W, X and Y are variables, X contains no Neg, and Neg is exhaustively dominated by a VP Rule (4.118) is restricted to a small class of verbs, VTh’ such as think, suppose, believe, guggg and Kant. Note the condition "X contains no Neg," which is necessary to block 332 the generation of ungrammatical sentences as in: (4.119) [[IJNP [pres]Aux [[ngtheg [think]v [[[heJNP [will]Aux [read a paper]VP [there]AdVL [EEQEJNngVPJSZJNPJVPJSl'IT’ *I don;t n93 think that he will read a paper there. Moreover, we note that though (4.118) is quite similar to Negative-raising, the relevant difference is that (4.118) operates on the negatives which are exhaustively dominated by a verb phrase, thus limiting itself to a subclass of negatives, sentential negatives in our analysis. To illustrate the difference with an example, consider the following: (4.120) a. I believe that he should not call her today. b. I donLt believe that he should call her today. If Negative-raising is an optional rule, there is no reason to block its application in this case. Thus, its optional application to the structure underlying (4.120a) gives sentence (4.120b). But the trouble with this analysis is that sentence (4.120b) is not synonymous with (4.120a). In contrast, our rule (4.118) cannot so apply to derive (4.120b) because the negative in (4.120a) is not exhaustively dominated by a verb phrase, as shown in (4.121) underlying (4.120a): 333 .121) s (4 ,'v"’:::::::’:h\\ NP Aux VP N pies V’////’ \\\\\NP i believed”:::::;,g2 NP Aux \:VP\\\\\\“AdvT N should Neg I NP today he not call N her Under our analysis, the negative in (4.121) is verb-phrase negation and only sentential negation, exhaustively dominated by a verb phrase, may undergo rule (4.118). Considering rule (4.118), we note its striking simi- larity in Operation to Negative-attachment in (3.221), especially (3.22lb): both of them attach a sentential nega- tive to a verb. Thus, we may conjoin them into: (4.122) a.(=3.221a) Quant Ava X Adv? Y Neg -—9 Ade Quant Ava X Neg+ AdvF Y Ade b.(=3.221b) v Y Neg-——9 X X Neg+V Y C.(=4.118) X VTh [ W Neg 38 Y -—9 X Neg+VTh [ W 38 Y where W, X and Y are variables, Neg is 334 exhaustively dominated by a VP, and X contains no Neg, Quant, AdvF, Ava such as gnly, AdvR-P or Ade The conjoinability of (4.118) with Negative-attachment means that in our analysis a minor rule, similar to Negative-raising, can be incorporated into a more general rule, Negative-attachment. We may retain the name Negative- attachment for rule (4.122). Now, let us consider the synonymity of (4.111) with (4.112). In fact, the synonymity of (4.111) and (4.112) can be accounted for in terms of the scope of negation which extends over the whole sentence in which it appears. Observe that the scope of negation in (4.114) underlying (4.111) is 82, and that in (4.115) underlying (4.112) is also 82: (4.123) [[he]NP [will]Aux [[not]Neg [readJV [a paperJNPJVP [therelAdeJS2 (4.124) [[[[he]NP [will]Aux [[read]V [a paperJNP]VP [there] 1 I [[not] Ade S3 NP N 2 The relevant parts included in the scope of negation in ngVPJS (4.123) are the same as in (4.124). Accordingly, sentences (4.111) and (4.112) are as synonymous as are sentences (4.125) and (4.126) derived from (4.123) and (4.124), respectively: (4.125) He will n23 read a paper there. (4.126) It is not so that he will read a paper there. 335 The synonymity of (4.125) and (4.126) may be tested in a number of ways. First, if we conjoin them with that lfi: we get a grammatical sentence as in: (4.127) a. He will not read a paper there, that is, it is not so that he will read a paper there. b. It is not so that he will read a paper there, that t§, he will not read a paper there. Second, if we negate one of them and conjoin it with the other in terms of ttgt lg, we get an ungrammatical sentence as in the following: (4.128) a. *He will read a paper there, that is, it is not so that he will read a paper there. b. *It is so that he will read a paper there, that lg, he will not read a paper there. Third, any of the following sentences may occur following both (4.125) and (4.126): (4.129) a. but she will b. but he will write a paper 0. but he will read a magazine d. but he will read a paper here This behavior of (4.125) and (4.126) indicates their synonymity. To confirm the point, we note that a synonymous pair of sentences of other types, for instance, active and passive sentences behave quite similarly to sentences (4.125) and (4.126). To mention just one example, consider the following: (4.130) They did not murder the pickpocket. 336 (4.131) The pickpocket was not murdered by them. First, conjoin them in terms of that ta as in: (4.132) a. They did not murder the pickpocket, that ta, the pickpocket was not murdered by them. b. The pickpocket was not murdered by them, that ta, they did not murder the pickpocket. We observe that the sentences of (4.132) are grammatical. Second, negating one of them and conjoining it with the other, we get an ungrammatical sentence in either case: (4.133) a. *They murdered the pickpocket, that is, the pickpocket was not murdered by them. b. *The pickpocket was murdered by them, lfléfi.l§9 they did not murder the pickpocket. Third, any of the following sentences may occur following both (4.130) and (4.131): (4.134) a. but you murdered him a’. but he was murdered by you b. but they examined him b’. but he was examined by them 0. but they murdered the burglar c . but the burglar was murdered by them Thus, if sentences (4.130) and (4.131) are synonymous, then (4.125) and (4.126) are also synonymous on the same ground. If (4.125) and (4.126) are synonymous, then it follows that (4.111) and (4.112) are synonymous. Summarizing so far, we have demonstrated that our analysis can account for the synonymity of sentences (4.111) 337 and (4.112) on the general principle that can also take care of the synonymity of sentences such as (4.125) and (4.126). 4.4. Advantages f the Proposed Analysis t_ Ehglish The discussion in the preceding section does not show clear advantages of our analysis over the rule of Negative- raising. In this section, however, we will demonstrate that the approach in terms of Negative—raising runs into dif- ficulties when extended. It is clear that if sentences of the type "NP think(s) [ X — not — Y JS" are not synonymous with their counterparts of the type "NP do(es) not think [ X - Y JS," they become counter-examples to Negative- raising. Although Negative—raising may take care of counter-examples of that kind by including a constraint, a minor rule with a constraint will be of little interest. Moreover, in such counter-examples, the approach in terms of Negative—raising will have to set up two distinct underlying structures for sentences of the two types in question, since they are not synonymous. Thus, the approach must distin- guish two cases, cases where sentences of the two types are synonymous and those where they are not synonymous. Clearly it will be preferable if we can treat them in a unitary way. In this section, we will cite a number of cases in which sentences of the two types are not synonymous, and demonstrate that in our analysis they can be explained in the same way as those cases where sentences of the two types 338 are synonymous. First Of all, we note that there are a number of cases in which sentences Of the two types are not synonymous. This is the case especially when they involve quantifiers and those adverbials discussed in Chapter III. Let us consider first the cases involving quantifiers. To take a concrete example, consider the following: (4.135) I think that all the critics did 22: like that movie. (4.136) I dohtt think that all the critics liked that movie. We observe that sentence (4.136) is not synonymous with (4.135). If we extend the analysis in terms of Negative- raising to this case, they will be derived from the same underlying structure: the Optional application of Negative- raising will give (4.136). The trouble with this analysis is obvious, since (4.135) and (4.136) are not synonymous. 0r otherwise, Negative-raising needs some constraint to block its application in cases such as this. Then, the approach in terms of Negative-raising will have to derive sentences (4.135) and (4.136) from different underlying structures. Thus, it fails to treat sentences such as (4.135) and (4.136) in the same way as sentences such as (4.111) and (4.112) in which the two types are synonymous. Under our analysis, on the other hand, sentences (4.111) and (4.112) are derived from different underlying structures, as discussed above: the former involves 339 verb-phrase negation, while the latter involves sentential negation. Extending this analysis, sentences (4.135) and (4.136) are derived from correspondingly different structures such as (4.137) involving verb-phrase negation and (4.138) involving sentential negation, respectively: Aux NW pres NP VP I I S3 Quant NP Aux all I ¢”/’;¥P\ Quant Det N past Neg NP / \ A the critics not like Det N that movie (4.138) ,,,/”:::::SI\\\\\ AIX \f’////VP\\\\\~ NP pres F4-4Z-% I think S I? VIP S3 Neg NP Aux VP not \‘\\\‘\. I ’/’ Quant Det N past V IP a l the critics like IDet \\\N that movie 340 Thus, the semantic difference between (4.135) and (4.136) is reduced to the relevant difference between (4.137) and (4.138): in (4.137) the sc0pe of the negative is S3 and the quantifier is outside Of its scope, whereas in (4.138) the negative, whose scope is 82, includes the quantifier in its scope. Now, turning to the derivation of (4.135) from (4.137), Sentence-raising applies to raise S3 into 82 and Quantifier-attachment assigns all to A, giving the intermediate structure (4.139): (4.139) ’/”/:;73 I\\\~W NP Aux I 1 VV/ P\ N pres t'P I think 82 NP /l W‘~\‘\“- Quarg Net\ N piast Ne/V} P\NP I I I I I / \ all the critics not like Det N I that movie The structure (4.139) generates sentence (4.135). Observing (4.139), we see that it satisfies the condition of Sentence-raising. If this transformation optionally applies to (4.139), the following intermediate structure is derived: 341 (4.140) s /’//////; l‘\\ NP Aux VP NP V N pres _l I did not like I think 'all the critics that movie Then, moving "I think" to follow the subject noun phrase, we get sentence (4.141), which is synonymous with (4.135): (4.141) All the critics, I think, did as: like that movie. On the other hand, in the case of (4.138), the appli- cation of Sentence-raising derives the intermediate structure: (4.142) 81 na””X§§::;; '::::VP\\\\\ v NP N pres I I M32 /“I’\ At >P\NI; Quant Det N past V //’ all tte critics li e Det I not I that movie 'Then, Negative-attachment (4.122c) applies to attach not to the verb think, giving (4.143): 342 (4.143) NP’77A;§;:::;;Sl\\\/\‘VP I /\\\\\‘ N pres /V NP I Neg/\ / S\2 nlot think NP Aux VP all the critics like that movie The structure (4.143) becomes sentence (4.136). Considering (4.142), we note that the negative hat can also be attached to the quantifier by the application Of Negative—attachment (4.122a), generating sentence (4.144), which is synonymous with (4.136): (4.144) I think that 22: all the critics liked that movie. In contrast, the attachment of hat to the verb ttha is blocked through the filtering function of Negative- attachment. Thus, we can block the following derivation: (4.145) [EIJNP [pres]Aux [[thinklv [[[éll the critics]NP [past]Aux [[likeJV [that movieJNP]VP NngVPJSZJNPJVPJSl--7L’ I think that at; the critics did hat like that [[ngt] movie. IFurthermore, we note that (4.142) meets the condition of ESentence-raising. The Optional application of this trans- :formation, after the attachment of not to all, gives (4.146): 343 (4.146) NP“'"XE§:::%§EEEEEEEEEEE§2:777‘777777777II“VP % Pies N irant//’§:t N' [///\\\\\k I think Neg t e critics liked that movie nét all Shifting "I think" to follow the subject noun phrase, we derive sentence (4.147), which is synonymous with (4.136): (4.147) Not all the critics, I think, liked that movie. Returning to (4.138), if Sentence-raising does not apply, the complementizer that, and ii: lfi and _a are inserted into S3 and 82, respectively, yielding sentence (4.148): (4.148) I think that it is 22: so that all the critics liked that movie. Thus, our analysis can account for the derivation of sentences (4.136), (4.144), (4.147) and (4.148) from the structure (4.138) as well as the derivation of sentences (4.135) and (4.141) from the structure (4.137) without resorting to any new transformation. By so doing, the non- synonymity Of (4.136) with (4.135) is automatically explained, since they are derived from the distinct underlying structures (4.138) and (4.137): hat in (4.138) includes all in its scope, while as: in (4.137) does not so include att. Next, let us turn to the cases involving negation and the adverbials cited in Chapter III, specifically, nominal 344 adverbials, adverbials of frequency and adverbials of reason and purpose. To start with, consider examples such as (4.149) and (4.150) involving a nominal adverbial: (4.149) I think that gnly that child did he: break the rule. (4.150) 1 don;t_think that only that child broke the rule. Clearly these two sentences are not synonymous, thus presenting another problem for Negative-raising. Within our framework, sentences such as (4.149) and (4.150) pose no problem: they are derived from different underlying structures such as (4.151) and (4.152), respectively: (4.151) ””’,;7Sl\\\\\\‘ NP Aux VP I ? Pies Y’//// \\\\NP I think 8 2 I{P/ \ V‘P S Adv NPvfl/y/ff”; i 3~‘\\\\\\‘VP oJI: ' //I ‘~\\\\~ ‘ //’//j/ \\\\\ A? N.Det N past Neg V l/NE\\ A that child not brLak Drt N the r e 345 (4.152) s //’l\ NP Aux VP I 1'/\ N pres V NP I think ’////,82\\\\\\ NP V? Neg H"fflf’f”’4;: 3\\\VP nLt /\ //,v Ava /Det\\\\‘N p;st V P\\\NP /\N only that child break Det the rule The structure (4.151) shows that the scope of the negative is S and the adverbial only is outside of its s00pe, 3 whereas the structure (4.152) indicates that the negative, whose sc0pe is 82, includes gnly in its scope. This relevant difference reflects the semantic difference between (4.149) and (4.150). Now, let us consider the derivation involving (4.151). Applying Sentence-raising and Quantifier-attachment to assign only to A, we derive the intermediate structure (4.153): 346 (4.153) a? Na 1 I ./”V W\\\\ N pres I Ava flt\\\\\N SZA/////////£\\\\\\\\\\‘ only that child did not break the rule This structure becomes sentence (4.149). Moreover, (4.153) meets the condition of Sentence—raising. Thus, its appli- cation gives (4.154): (4.154) A ’“33E;::::::; NP VP think only that child the rule Then, shifting "I think" to follow the subject noun phrase, we derive sentence (4.155): (4.155) Only that child, I think, did not break the rule. Turning next to (4.152), the application of Sentence- raising gives the intermediate structure (4.156): E? \% min-Ea // NP VP VP AdlvN/ D‘iat\1i1 /\ Ntleg only that child 'broke the rule not Then, if Negative-attachment (4.122a) applies to attach not to only, it gives sentence (4.157): (4.157) I think that not only that Child broke the rule. If, on the other hand, Negative-attachment (4.122c) applies, the intermediate structure (4.158) is derived: N? Aux \VP I ! /////’ ‘\\\\~ ? pres V h? I Neg S I / 2\ not think NP VP 44/’//\\\“\\i only that child broke the rule The structure (4.158) generates sentence (4.150). Moreover, Sentence—raising may further apply to (4.156), after the attachment of not to only, raising 32 into S1 to derive: Aux VP t ptes ( i//’/////A\\\\\\\\_ I think not only that child broke the rule (4.159) //s NP / l\NP\VP 348 Then, shifting "I think" to follow the subject noun phrase, we get sentence (4.160): (4.160) Ea: only that child, I think, broke the rule. Returning to (4.152) once again, if S3 is not raised into 32, the complementizer that, and it, is and so are inserted into S3 and 82, respectively, deriving sentence (4.161): (4.161) I think that it is not so that only that child broke the rule. In this way, our analysis can take care of the derivation of sentences (4.150), (4.157), (4.160) and (4.161) from (4.152) as well as that of sentences (4.149) and (4.155) from (4.151) without resorting to any new transformation. Next, let us consider sentences such as (4.162) and (4.165) involving an adverbial of frequency: (4.162) I think that John Qiian does he: do his best. (4.163) I don;t think that John giten does his best. Clearly (4.162) and (4.163) are not synonymous, thus providing still another problem for the approach in terms of Negative—raising. Within our framework, they will be derived from dif- ferent underlying structures such as (4.164) and (4.165), respectively, with unrelated details aside: 349 (4.164) //Sl\ NP Aux VP N pies Y’////’ \\\\\\?P I minim/SN VP ‘ l S Adv 3 F I 1 // \\\\ T pres Neg' VK PP John ntt do 4::f::>§ his best (4.165) /Sl\v NP Aux P ‘ ”//’ ‘\\\\~ N pres NP I think S / 2 \ NP VP 3‘ N‘ -eg NP WP not S Adv ’/,/” 4 F NP Au/x \VP 0 ten l I / N pres V L Jthn 0 his best The structure (4.164) indicates that the sc0pe of 11.9.: is 83 euui often is outside of its s00pe, whereas the structure (4.165) shows that the negative, whose scope is 82, includes EEEEQQ in its SCOpe. This difference accounts for the semantic difference between (4.162) and (4.163). 350 Now, let us consider the derivation involving (4.164). Applying Sentence—raising and Adverbial-movement, we derive sentence (4.162). If Sentense-raising further applies to raise 82 into 81’ it derives sentence (4.166): (4.166) John, I think, gttgg does Qgt do his best. 0n the other hand, if Sentence—raising does not apply to raise S3 into S2, the complementizer that and the copula t§ are inserted, yielding the following sentence: (4.167) I think that it is gflen that John does ngt do his best. In the case of (4.165), applying Sentence—raising to raise S4 into 83, Adverbial-movement and Sentence-raising to raise S3 into 82, we get (4.168): (“68) 4/77314\ ux 1 | / N pres V ‘\\\NP 1 NP Aux Adv;/' VP VP I l I / \ I N pres often Y NP Nfg John do not his best Then, the application of Negative—attachment (4.122c) gives sentence (4.163). If, on the other hand, Negative- attachment (4.122a) applies to (4.168), attaching ggt to Qttgg, it yields sentence (4.169): (4.169) I think that John does not often do his best. 351 After the application of Negative-attachment, if 82 is further raised up into 31’ it derives sentence (4.170): (4.170) John, I think, does not often do his best. Let us return to (4.165) once again. After S4 is raised into 83, if the adverbial is not shifted but S3 is further raised into S2, the resulting structure is as follows: (4.171)"”#”::::;,sl\\ NP Aux VP l ,/”’ \\\\ NP N pres V I I think S NP Aux VP ‘\\\‘“VP VP l l x/' \\ l I N pres V NP AdlvF Neg John do often not his best Then, Negative-attachment (4.122b) applies to attach npt to the verb dg, giving sentence (4.172): (4.172) I think that John does npt do his best gttgn. Furthermore, if 83 in (4.165) is not raised into 82, we get sentence (4.173): (4.173) I think that it is pgt so that John gttgg does his best. Thus, our analysis can derive the synonymous sentences (4.163), (4.169), (4.170), (4.172) and (4.173) from the ‘underlying structure (4.165) as well as generate the synonymous sentences (4.162), (4.166) and (4.167) from the 352 structure (4.164) without involving any new rule. Next, let us consider sentences such as (4.174) and (4.175) involving an adverbial of reason: (4.174) I believe that the president will pgt employ you for that reason. (4.175) I dontt believe that the president will employ you for that reason. At first sight, it seems that the Negative-raising analysis can relate the sentences by deriving them from the same underlying structure: the Optional application of Negative- raising gives sentence (4.175). 0n closer examination, however, we note that (4.174) may be ambiguous, while (4.175) is unambiguous. This clearly poses another problem for the rule of Negative-raising. Furthermore, consider the following: (4.176) I believe that for that reason the president will ggt employ you. (4.177) I dontt believe that for that reason the president will employ you. Sentence (4.176) is clearly not synonymous with (4.177), which presents still another problem for the Negative- raising rule. Under our analysis, since they are not synonymous, they are derived from different underlying structures such as (4.178) and (4.179), reSpectively: 353 (4.178) s .//‘ l\ NP Aux .VP N pres N NP I I believe”””///,,.SZ~\\-‘\\\\\‘~ 1)? VF Ad NP Aux P / \ I /I \ Bit N will Nfg N NP for that reason the president not employ N you (4.179) s NP Aux VP N pres V NP I believe S IIJP/ 2\:\:‘P S 8% NP/ 3 \VP ‘ ' not Adv NP E\\\ / \ \ / I Det N will V NP for that reasofi' the president employ N you {The relevant difference between them is that in (4.178) the :SCOpe of the negative is 83 and the adverbial is outside of ;its scope, whereas in (4.179) the negative, whose scope is :32, includes the adverbial in its scope. This difference 354 reflects the semantic difference between (4.176) and (4.177). I Now, let us consider the derivation involving (4.178). Applying Sentence—raising to raise S3 into 82, we get the intermediate structure (4.180): (4.180) /AuX//S /1\V pies V///V P\\\\NP believe /PNP\ VP‘ étw N will Neg /V Adv I I I I R he president not employ N D t you for that reason The structure (4.180) becomes sentence (4.174) with a reading synonymous with (4.176). If the adverbial in (4.180) is optionally shifted to precede the subject noun phrase, it gives sentence (4.176). Moreover, if Sentence-raising further applies to raise 82 into 81’ it yields sentence (4.181): (4.181) For that reason, I believe, the president will Qgt employ you. In the case of (4.179), on the other hand, applying Sentence-raising twice to raise 84 into S3 and 83 into 82, we get the intermediate structure (4.182): 355 (4.182) s NP Aux VP N pres V NP I A I believe NF"””'.X;;::::::::35’é::::::::::v§““‘vr Det/\ 1 )\ I ‘ N will Y NP AdvR Neg the president employ N not you for that reason Then, applying Negative-attachment (4.122b), we get sentence (4.174) with a reading synonymous with (4.177). This provides an explanation for the ambiguity of sentence (4.174): it is derived from (4.179) as well as (4.178). What is more, if S2 in (4.182) is Optionally raised into S1, it gives sentence (4.183): (4.183) The president, I believe, will not employ you for that reason. This sentence is also ambiguous. When it has another reading synonymous with (4.176), it is derived from (4.184), which is in turn derived from (4.180) by raising 82 into 81: (4.184) s NP Aui”IV§:::::::;g’ ‘:A§;‘\\“‘VP VP % pies 4 4///\\\\_will 1////\\\\; AJV I the not employ “R I believe president you for that reason Neving I believe to follow the subject noun phrase, we get sentence (4.183) with a reading synonymous with (4.176). 356 Now, let us return to (4.179). After 84 is raised into 85’ if the adverbial of reason is optionally shifted to precede the subject noun phrase, and 83 is further raised into S2, the resulting structure is as follows: (4.185;?Lfl7:12;:::::;—;13l\\V V./”””V P\\\‘\~m beli::j,,»”””’7”.fg:\~““‘-;~1~‘_‘v dVR et/NP\ I V/VP\ VI I I N pres I D N will NP Neg I I president employ N not I for that reason t you Observing (4.185), we note that Negative-attachment (4.122b) cannot apply to attach not to employ, since the adverbial of reason precedes the verb employ. But, Negative—attachment (4.122c) can apply, attaching not to believe to derive (4.186): (4.186) NP//Sl\, ATX I’,,,r/' P~\\\\\\\‘NP I ”‘7 I Neg not believe Advgi’III7‘S 2.\“;:;:N“““VP R/\ I VV/\NP Det N will I for that reason the president employ N you 357 The structure (4.186) generates sentence (4.177). Thus, our analysis can explain the semantic difference between (4.174) and (4.175). and between (4.176) and (4.177). In addition, this analysis can account for the ambiguity of sentences such as (4.174) and (4.183). This analysis can be extended to the cases involving adverbials of purpose including benefactive adverbials. In addition to them, we may cite the following sentences involving a modal auxiliary: (4.187) I believe that he must pg: call her today. (4.188) I donit believe that he mpg; call her today. Clearly (4.187) is not synonymous with (4.188), thus presenting another problem for Negative-raising. Under our analysis, however, they present no problem: they are derived from different underlying structures such as (4.189) and (4.190), respectively: (4.189) //S1\ 358 (4.190) s N§77723;;;, l\\\‘VP I l V’,//’/’ \\\\\NP N pres I 1 I I be ieve ’,,/”’782‘N\‘~\‘ NP VP S 98 /3\ NP Aux VP AdvT not ,/’ \x l N must Y NP today he call N her It should be noted that in (4.189) the negative, which is not exhaustively dominated by a verb phrase, does not meet the condition of Negative-attachment. Thus, there is no chance of n93 in (4.189) being attached to believe to derive (4.188). On the other hand, the structure (4.190) indicates that the negative meets the condition of Negative- attachment, thus yielding sentence (4.188). It also generates the following sentence: (4.191) I believe that it is not so that he mugt call her today. Sentence (4.191), though a little wordy, is synonymous with (4.188), not with (4.187). Quite similarly, the non- synonymity of (4.192) with (4.193) may be explained in our analysis: should not (4'192) I believe that She'iought not to I attend the ceremony. 559 attend (4°193) I don't believe that she (should } ought 3Q the ceremony. Expanding this discussion, we may now go on to consider examples such as (4.194) and (4.195): (4.194) I think that nobody will take care of the old man. (4.195) I don;y think that anybody will take care of the old man. We have already discussed in section 2.8 that the embedded sentence of (4.194) contains sentential negation rather than verb-phrase negation. Thus, both (4.194) and (4.195) are derived from the same underlying structure: (4.196)"”#””,,14S /1\ NP Aux VP N pres NP I NP VP 4 N' e8 Aux VP not I Quant N will 84y bo y take care of the old man If S3 is not raised into 82, we get sentence (4.197): (4.197) I think that it is p93 so that anybody will take care of the old man. If, on the other hand, S3 is raised into 82, it derives the intermediate structure (4.198): 360 (4.198) s A// 1\ Quant N w111 Neg [ I take care of I any body the old man not Then, Negative-attachment (4.122a) applies to attach not to any, deriving sentence (4.199): (4.199) I think that not anybody will take care of the old man. Next, n93 is incorporated into anybody, yielding sentence (4.194). Furthermore, Negative-attachment (4.122c) can also apply to (4.198), attaching not to think to derive: A/ux/'Sl\VP I v / \NP pres (4.200) IH——2h—E§ I Neg S not think NP Aux | P A /\ take care of any-Ody the old man The structure (4.200) becomes sentence (4.195). Incidentally, notice that Negative—attachment (4.122b) cannot apply to (4.198), since the quantifier any precedes the verb take. Thus, it blocks the following derivation: 361 (4.201) [[IJNP [pres]Aux [[thinkJV [[[anybodyJNP [will]Aux [take care of the 01d man]VP [[EgijNngVPJSZJNPJVPJSl'7*’ *I think that anybody will go: take care of the old man. In this way, examples (4.194) and (4.195) provide additional confirmation for the existence of sentential negation opposed to verb-phrase negation as well as show the validity of the analysis in this section. In this connection, the following example may be cited from Jackendoff (1971, 290): (4.202) John doesnLt think that 3111 didnLt go. Jackendoff argues that if (4.202) is derived by the appli- cation of Negative—raising, the source of the commital sense of (4.202) has to be an ungrammatical sentence as follows: (4.203) *John thinks that Bill didnLi not go. He argues that as Negative-raising is an optional rule, one would expect (4.205) to be grammatical. Now, let us consider (4.202) within our framework. Sentence (4.202) is derived, under our analysis, not from (4.203) but from (4.204): 362 (4.204) s ””,,,:::77 l\\\\\ NP Aux VP I I ’/’//’ ‘\\\\\ N pres Y NP JLhn think ’//’//,S2‘\\\\‘ NP VP s‘, N‘ 98 liIP A1|1x /V P\V not N past Neg Bill not go It should be noted that (4.204) contains both sentential and verb—phrase negation. If 33 of (4.204) is raised into S the resulting structure is: 2’ (4.205) ’,,,””’/’Sl NP Aux/ \VP N | V//’///’V P\\\\\m I pres I I? John think S / 2 NP Aux/ \VP\ VP I I \ I N past Nfg V Nfg Bill not go not Observing (4.205), we note that the sentential negative n9: cannot be attached to the verb g9, since the occurrence of another negative preceding g9 violates the condition of Negative-attachment (4.122b). Thus, the generation of sentence (4.205) is correctly blocked in our analysis. Furthermore, Negative-attachment (4.122c) can apply to (4.205), attaching the sentential negative not to think to derive (4.206): (4.206) //Sl\ NP Aux VP N pres V NP John Neg S ! /// 2\\\v not think NP P N 1 Bill ‘did not“go The structure (4.206) generates sentence (4.202). Returning to (4.204) once again, if S3 is not raised into 82, it derives sentence (4.207): (4.207) John thinks that it is not so that Bill did 221 so. Sentence (4.207), though wordy, is synonymous with (4.202). In this way, example (4.202) further confirms the validity of our analysis involving the twofold distinction of negation. Amplifying the above argument, let us consider (4.202) from a different viewpoint. If Negative—raising is an Optional rule, there is no reason to block its application to derive (4.208) from the structure underlying (4.202), unless some ad hgg constraint is imposed on the rule to block it: (4.208) *John doesn;t asi think that Bill went. This problem does not occur within our framework. Negative- attachment to attach a negative to a higher verb is 564 restricted to sentential negatives, namely, negatives exhaustively dominated by a verb phrase, as stated in (4.122). Under our analysis, the structure (4.204) underlies (4.202) and the second negative in (4.204) cannot be attached to the higher verb yhink, since it is verb- phrase negation, not exhaustively dominated by a verb phrase. Furthermore, we note that sentence (4.202) has another reading and in that case it is derived from (4.209) under our analysis: NP Aux VP I / N pies Neg////’V \\\\\‘NP I I I I John not think S2 NP Aux VP I / \ N past Neg V Bill not gL The structure (4.209) indicates that the negative in 82, which is not exhaustively dominated by a verb phrase, cannot undergo the Negative—attachment rule. Therefore, there is no chance of sentence (4.208) being generated from (4.209). Thus, our analysis can block the generation of ungrammatical sentences such as (4.205) and (4.208), in addition to deriving grammatical sentences such as (4.202). 365 4.5. Conclusions From the discussion in this chapter, we may conclude that: 1. Sentences of the type "NP think(s) [ X - not - Y 33" and their counterparts of the type "NP do(es) not think [ X - Y JS" are derived from different underlying structures: the former involves verb—phrase negation,v while the latter involves sentential negation. The synonymity or the non-synonymity of sentences of the two types in question, as the case may be, can be accounted for in terms of the presumably universal constraint: the scope of negation, sentential or verb— phrase, is the whole sentence in which it occurs. The advantage of our analysis is that it can take care of a number of cases where sentences of the two types are different in meaning or grammaticality in the same way as those cases where sentences of the two types are synonymous: one type involves verb-phrase negation and the other type involves sentential negation, regardless of whether they are synonymous or not. This constitutes another motivation for the twofold distinction of negation. Another advantage is that in our analysis the derivation of sentences of these two types from their distinct underlying structures involves no new transformation except those which are shown to be independently motivated in Chapter II and III, such as 566 Sentence-raising, Negative-attachment and Quantifier-attachment. Thus, our analysis can treat them in a relatively general way. In particular, a minor rule such as Negative-raising can be incorporated into the general rule of Negative-attachment in our analysis. The minor rule of Negative-raising, therefore, can be eliminated from both Japanese and English grammars. The analysis prOposed in Chapter II has its validity further confirmed in that it can take care of those facts regarding the Negative-raising rule. Furthermore, the proposed analysis has several advantages over the rule of Negative-raising in that it can account for a number of other cases that cannot be explained in terms of Negative-raising. CHAPTER IV FOOTNOTES 1. Negative-raising or Negative transportation is a rule to relate sentences such as: (a) I think that Bill doesn't like Mary. (b) I don't think that Bill likes Mary. If Negative-raising optionally applies to the structure underlying (a), it gives sentence (b). 2. This sentence may be ambiguous. When it has another reading in which the verb omou "think" is negated, it is derived from a different underlying structure such as: (a) ‘flfflffl”fyyfl,,iaes5:::::~y~“-‘-‘--~ NP VP ij N NP N fg pres watasi ga S omou nai 2 I NP’/’z/Iizfltp\\\\\“ think not Aux I ,/” N NP V pres John ga N syootaisuru Mary 0 inv1te The negation in (a) is verb-phrase negation which negates the higher verb omou "think." Throughout this chapter, we ignore, as irrelevant, the readings of the 567 368 (a) type in which the higher verb omou "think” is negated. Notice that there is no correspondent of sika in English since "sika . . . nai" as a unit corresponds to only, as noted in section 2.1. Japanese distinguishes the verb gmgg from omotteiru in the present tense: gmgg may be used only when the subject is in the first person, as opposed to omotteiru which may occur regardless of the person of the subject. Thus, the following correSpondence may be observed between the Japanese omou/omotteiru and the English yhigk in the present tense: (a) 1. Japanese: watasi wa X to gmgg / omowa nai English: I :hink / don't think that X ii. Japanese: watasi wa X to omotteiru / omotte inai English: I think / don't think that X (b) Japanese: NP wa X to omotteiru / omotte inai English: NP think(s) / do(es) not think that X where NP is not in the first person Thus, when the subject is in the first person, both omou and omotteiru may be used but with a different meaning. For instance, compare the following examples: 569 (c) (watasi wa) Bill ga kokoni ko nai to omou. I here come not that think 'I think that Bill will not come here.‘ (d) watasi wa Bill ga kokoni ko nai to omotteiru. I here come not that think 'I think that Bill will not come here.‘ We observe first that watasi yg may be optionally deleted in (0) but not in (d). Moreover, there is some semantic difference between (0) and (d): watasi yg gmgg in (0) often expresses the probability of the contents of the embedded sentence, viewed from the standpoint of the speaker, so sentence (c) is almost synonymous with: (e) tabun Bill wa kokoni ko nai daroo. probably here come not will 'Probably Bill will not come here.‘ On the other hand, sentence (d) asserts or focuses the speaker's mental action of thinking, so it is in the same class with sentences such as (f) and (g) in which the subject is in the second or third person: (f) anata yg Bill ga kokoni ko nai to omotteiru. you here come not that think 'You think that Bill will not come here.‘ (g) kanozyo ya Bill ga kokoni ko nai to omotteiru. she here come not that think 'She thinks that Bill will not come here.’ Furthermore, if soo fig g3 "is it 50?", a kind of tag question, is attached to (c) and (d), there is some difference in grammaticality between the resulting 370 sentences (h) and (i): (h) (watasi wa) Bill ga kokoni ko nai to omou, I here come not that think soo g; gg. so is Q(uestion) M(arker) 'I think that Bill will not come here, is it so?‘ (i) ?watasi wa Bill ga kokoni k0 nai to omotteiru, I here come not that think see as as. so is QM 'I think that Bill will not come here, is it so?’ We observe that (h) is grammatical in Japanese, while (i) does not sound as grammatical as (h). In this connection, consider the following: (j) tabun Bill wa kokoni k0 nai daroo, soo fig ng. probably here come not will so is QM 'Lit. Probably Bill will not come here, is it so?‘ As noted above, sentence (e) is almost synonymous with (0), so (j) is almost synonymous with (h). Then, it follows that as ggg dg ng "is it so?" in (j) refers to "Bill wa kokoni k0 nai daroo (2 Bill will not come here)," so ggg gg _g "is it so?" in (h) refers to the embedded sentence "Bill wa kokoni ko nai (= Lit. Bill does not come here)." This fact makes sentence (h) acceptable, since it does not violate the constraint, pointed out by R. Lakoff (1969a, 145), that for verbs of mental action such as think it is impossible for the subject to ask whether they are true of him. 371 In contrast, sentence (i) is in the same class with (k) and (l): (k) anata wa Bill ga kokoni k0 nai to omotteiru, you here come not that think ___soo 9.2 as. so is QM 'You think that Bill will not come here, is it so?‘ (1) kanozyo wa Bill ga kokoni k0 nai to omotteiru, she here come not that think __800 is as. so is QM 'She thinks that Bill will not come here, is it so?‘ As soo d e "is it so?" in (k) and (1) cannot refer to the embedded sentence but to the matrix sentence, so U) _pp fig _g in (i) usually cannot refer to the embedded sentence but to the matrix sentence. Then (i) is a sentence in which the speaker asks whether his own mental action is true or not, thus violating R. Lakoff's constraint. In this way, the difference in grammati- cality between (h) and (i) is explained in terms of R. Lakoff's constraint. Yet, (i) does not sound completely ungrammatical. This seems to be partly due to the analogy of (h). To take this argument one step further, consider the following examples, cited from R. Lakoff (1969a): (m) I suppose you think you're real smart, don't you? (n) I suppose John isn't here, lg pg? 372 (o) I don't suppose the Yankees will win, will they? Assuming that these sentences are acceptable, as R. Lakoff says, we note that they involve the subject in the first person and suppose in the present tense. Moreover, their tag questions refer to the embedded sentences, not the matrix sentences. Accordingly, I suppose in (m) and (n) corresponds to watasi yg gggg, not watasi yg omotteiru in Japanese, that is, I suppose in (m) and (n) belongs to (a.i), not (a.ii). Similarly, I don't suppose in (o) correSponds to watasi yg gggyg pg; rather than watasi yg omotte inai in Japanese, thus belonging to (a.i), not (a.ii). Summarizing the discussion, we see that when the tense is in the present and the subject is in the first person, Japanese distinguishes gggg and omotteiru as in (a.i) and (a.ii) and that only (a.i) may have a tag question attached to it. Correspondingly, though English makes no apparent distinction between (a.i) and (a.ii), only the English equivalent of (a.i) may have a tag question attached to it, as shown by the grammaticality of (m), (n) and (o). Incidentally, Japanese has another adverbial hotondo that is the homonym of the hotondo in question here. This homonym means "almost," "nearly" or the like and it may occur in affirmative environments, as follows: 373 (a) tatemono wa hotondo kanseisita. building almost completed 'The building is almost completed.’ Therefore, sentence (b) may be grammatical if hotondo in (b) is interpreted as meaning "almost, nearly" by the help of the accompanying context, as in (c): (b) musuko wa hotondo benkyoosuru. son study (0) itinitizyuu ie ni tozikomotte musuko yg all day long house in stay son hotondo benkyoosuru. almost study 'My son stays indoors all day long and studies almost all the while.‘ We notice that sentence (b) is not exactly complete by itself but some element understood to be modified by hotondo is deleted. For instance, sentence (b) may be synonymous with either of the following, according to the context in which it occurs: (d) i. musuko wa sono aida hotondo benkyoosuru. son the while almost study 'My son studies almost all the while.‘ ii. musuko wa hotondo itinitizyuu benkyoosuru. son almost all day long study 'My son studies almost all day long.‘ iii. musuko wa hotondo mainiti benkyoosuru. son almost every day study 'My son studies almost every day.‘ On the other hand, sentences like the embedded sentence 374 of (4.90b) involving hotondo of another kind are complete by themselves and no element to be modified by hotondo can be inserted in any context. If the literal translation of (4.107) is grammatical, this is partly due to the difference between the Japanese gggg and its English equivalent ggpII 0T.£lll- Moreover, it seems to be partly due to the word order made . . . nai in Japanese and not . . . until/till in English; notice that in (4.107) Eggg precedes the negative ggI, while in its English translation the negative precedes gppII. Thus, as far as the word order is concerned, a more exact counterpart of (4.107) will be as in the following: (a) *My brother got up until nine, I don't think. This sentence may be ambiguous. When it has another reading in which the verb think is negated, it is derived from a structure of the following form: (a) S //l E? ATX :::::¥P I JP pres Neg V \\\\\\N I not think 82 NP Aux VP Adv L I I /’//I \\\ I N will V - NP there I he read a paper 375 The negation in (a) is verb-phrase negation that negates the higher verb nggg. Throughout the subsequent sections, we ignore, as irrelevant, the readings of the (a) type in which the higher verb nggk is negated. Incidentally, as for the ambiguity, R. Lakoff (1969a, 146) notes as follows: There is another interpretation, of course, with the higher verb itself being negated. Thus, "John doesn't think Bill likes Harriet" might have two interpretations: (1) "John thinks Bill doesn't like Harriet" (John has a definite Opinion): (2) "It isn't so that John thinks Bill likes Harriet" (John need not have any opinion; he might, in fact, not know anything about either Bill or Harriet or the feelings of the former for the latter. In this case, of course, negative—transportation has not occurred. [sic.] CHAPTER V CONCLUSIONS This thesis has been an attempt to formulate an analysis of negation that is applicable to both Japanese and English. The major findings of the thesis may be summarized as follows: 1. The proposed analysis involves the following base rules in Japanese and English: (5.1) Japanese: a. s -9'NP VP (Aux) (wa) b. ((Ade) (NP) V (Neg) Neg VP -—9 1 Quant Ava AdvF \AdVR-P (S) (Ava) (Quant) (Det) N “P “ ”Is (NP) I (5.2) English: C. a. S ‘——>NP (Aux) VP b. ((Neg) v (NP) (Ade)w Neg VP “‘9 II Quant r Ava I AdvF ) \AdvR.P 376 377 c. (Ava) (Quant) (Det) N (8)} NP‘—’IINP) 8 These rules are supported and justified by the putative universals concerning the scope of negation and that of quantifiers. In addition, this analysis involves the following transformational rules which are shown to be inde- pendently motivated in Japanese and English: (5.5) Japanese: a. Sentence-raising OPT [MEXJS NJS --)[M X N38 2 l 1 where M, N and X are variables, and M or N dominates VSoT or exhaustively dominates AdjS, Adv AdvF, AdVR.P' Quant or Neg N’ b. Contrastive yg—attachment Quant Ade X [Ade . . . VJVP ' Y wa Z —-9 AdvD . [AdvD . . . AdJJVP Quant+wa - Ade+wa X [Ade . . . Adv +wa D [AdvD . . . Adj+wa]VP V+wa]VP Y Z where X, Y and Z are variables 378 c. Negative-attachment i. {V} P Y Adj + wa X V Z Neg'-—€> Q + wa Y {Adj } V P Y IAdj + wa + Neg X { V Z Q + wa Y Adj ) + Neg ii. X [ W Neg 38 Y VTh Z-——) X [ W 33 Y VTh+Neg Z where P, Q, W, X, Y and Z are variables, Z contains no Neg, P includes Ade or AdvD, Q includes Quant, Ade or AdvD, and Neg is exhaustively dominated by a VP d. Quantifier-attachment Quant X [A (Det) N JNP Y {Adv }-—§ N Quant x [ (Ava } (Det) N JNP Y where X and Y are variables, and X contains no Quant, Ava or Neg e. Adverbial—movement [ X NP Y i Y AdVR BI AdvF Adv: “Adv AdvF Hi IAdv: ;} where X and Y are variables, and NP is immediately dominated by S 379 (5.4) English: a.(=5.5a) Sentence-raising OPT [ Mif X J N'] '——? [ M X N J S S S 2 l l where M, N and X are variables, and M or N dominates VS-T or exhaustively dominates AdjS, Adv Adv AdvR P’ Quant or Neg N’ F’ b. Negative-attachment i. Quant Ava X AdvF Y Neg -+ Ade Quant Ava X Neg+ AdVF Y Ade ii. X. V Y Neg -9 X Neg+V Y iii. X VTh [ W Neg 38 Y -—§ X Neg+VTh I W 38 Y where W, X and Y are variables, Neg is exhaustively dominated by a VP, and X contains no Neg, Quant, AdvF, Ava such as only, AdvR.P or Ade 580 c.(=5.5d) Quantifier-attachment Quant X [A (Det) NJNP Y IAvaI—3 {Quant X [ Adv I (Bet) N JNP Y where X and Y are variables, and X contains no Quant, Ava or Neg d. Adverbial—movement fl AdvF [ X NP Aux Y Adv } JS'—-9 R P NP Aux AdvF X [ AdvF Y IAdV' I NP Aux R P where X and Y are variables, and NP is immediately dominated by S It should be noted that Sentence-raising and Quantifier- attachment have the same form and are applicable to both Japanese and English. 2. The prOposed analysis incorporates, among others, the putative universals we noted regarding the scope of negation and that of quantifiers: (5.5) a) The scope of negation, sentential or verb- phrase, is the whole sentence in which it occurs. b) The scope of a quantifier, sentential or nominal, is the whole sentence in which it occurs. 5. From the putative universals it follows that if a 381 constituent is not included in the scope of negation or a quantifier, it must be outside the sentence containing the negation or the quantifier in the underlying structure. We noted that in sentences of the (5.6) type involving a negative and a quantifier in a simplex sentence there are only two possibilities: either Quant is included in the scope of Neg or not: (5.6) a. b. S Neg-f;;;;t::>- ¢5f/0;;nt:::Neg:>> More specifically, when Quant is included in the scope of Ne , Quant cannot include Neg in its sc0pe. Thus, the structure underlying (5.6) in that case is: S2 4am: The structure indicates that Neg, whose SCOpe is S Neg 1, includes Quant in its scope. Furthermore, (5.7) indicates that since the scope of ngpi is S2, Ngg is outside of its scope. In turn, when Qgggp in (5.6) is not included in the scope of Neg, Quant includes Egg in its scope. Thus, the structure underlying (5.6) in this case is: 382 (5.8) /Sl\ 82 Quant éNeg: The structure (5.8) indicates that Qggpp, whose SCOpe is Sl’ includes Ngg in its SCOpe and that since the scope of Ngg is 82, Qgggp is outside of its sc0pe. These considerations led us to make a twofold distinction for negation and quantifiers: as Ngg in (5.7) is sentential but Egg in (5.8) is verb-phrase, so Qgggp in (5.8) is sentential but Qgggp in (5.7) is nominal. Only senten— tial Egg and Qgggp, "commanding" the sentences they modify, can include another Qgggp or Egg in their scope. Next, noting that three classes of adverbials, nominal adverbials, adverbials of frequency and adverbials of reason and purpose, behave quite similarly to quanti— fiers with respect to negation, we presented a similar argument for them. For instance, in sentences of the (5.9) type, containing negation and a nominal adverbial in a simplex sentence, if ggyN is not included in the scope of Ngg, it must be outside the sentence containing Ngg in the underlying structure, since the scope of Ngg is the whole sentence in which it occurs: S b. /\ Ava-—:N;;::> Neg-—-Ava Thus, the structure underlying (5.9) is: (5.9) a. 383 (5.10). /1\ 82 Ava 4N9 g> The structure (5.10) indicates that since the SCOpe of Neg is 82, Ava is outside of its sc0pe. On the other hand, when Ava in (5.9) is included in the scope of Neg, it is synonymous with the sentence derived from the underlying structure of the (5.11) type: (5.11) 31 / \ 82 Neg 45::fAdv;::>; The structure (5.11) indicates that Neg, whose scope is 81’ includes Ava in its scope. Based on these consid- erations, we made a twofold distinction for Ava, similar to that for Qggpp: only sentential ggyN as in (5.10) can include negation in its scope. _ We also demonstrated that a similar analysis holds for adverbials of frequency and adverbials of reason and purpose in that, depending upon whether they occur with sentential or verb-phrase negation, they show similar semantic differences. The discussion of manner adverbials in Chapter III presents the following constraint which is another candidate for a language universal: 384 (5.12) Manner adverbials cannot co-occur with verb-phrase negation. The constraint involves the distinction between sentential and verb-phrase negation, thus providing further motivation for the twofold distinction of negation. The proposed analysis can also take care of a number of facts regarding the rule of Negative—raising. Under our analysis, sentences of the type "NP think(s) [ X - not - Y IS" and their counterparts of the type "NP do(es) not think [ X - Y JS" are derived from dif- ferent underlying structures: the former type involves verb-phrase negation, while the latter involves sentential negation, regardless of whether they are synonymous or not. The synonymity or the non—synonymity of sentences of the two types, as the case may be, can be predicted in terms of the presumably universal constraint: the s00pe of negation, sentential or verb—phrase, is the whole sentence in which it occurs. This additionally confirms the validity of the constraint. The advantage of the proposed analysis is that it can account for a number of cases that cannot be explained in terms of Negative-raising, namely, those cases where sentences of the two types are different in meaning or grammaticality; it can explain them in the same way as those cases where sentences of the two types are 385 synonymous, in terms of the twofold distinction of negation. Another advantage is that in our analysis the derivation of sentences of the two types involves no new transformation except those which are shown to be independently motivated in Chapter II and III. In particular, the minor rule of Negative—raising can be incorporated into the general rule of Negative- attachment in our analysis. Thus, our analysis can treat sentences of the two types in a relatively general way. BIBLIOGRAPHY BIBLIOGRAPHY Akmajian, Adrian, 1970. Aspects g: the Grammar gI Focus Ig English. Unpublished doctoral dissertation, M.I.T. Austin, John L., 1962. How pg 2g Things with Words. Cambridge, Mass.: Harvard University Press. Bach, Emmon, 1967. "Have and fig in English Syntax," Language, 43. 462-485. , 1968. "Nouns and Noun Phrases." In Bach and Harms. 90—122. , and Robert Harms (eds.), 1968. Universals Ig Linguistic Theory. New York: Holt, Rinehart and Winston, Inc. Baker, C. Leroy, 1970a. "A Note on Scope of Quantifiers and Negation," Linguistic Inquipy, 1. 156-158. , 1970b. "Double Negatives," Linguistic Inquiry, 1. 169-186. Bierwisch, Manfred, 1968. "Two Critical Problems in Accent Rules," Journal g: Linguistics, 4. 175-178. Bloch, Bernard B., 1946a. "Studies in Colloquial Japanese I: Inflection," Journal gI the American Oriental Societ , 66. 97-109. , 1946b. "Studies in Colloquial Japanese II: Syntax," La ua e, 22. 200-248. , 1946c. "Studies in Colloquial Japanese III: Derivation of Inflected Words," Journal gI the American Oriental Society, 66. 504-315. Bolinger, Dwight L., 1967. "The Imperative in English," Ig Honor Roman Jakobson, I. The Hague: Mouton and Company. 535-562. Bresnan, Joan, 1970. "On Complementizers: Towards a Syntactic Theory of Complement Types," Foundations ‘gf Langgage, 6. 297-321. 586 387 Bresnan, Joan, 1971a. "Sentence Stress and Syntactic Transformations," Lagggage, 47. 257—281. 9 1971b. "On 'A Non-Source for Comparatives'," Linguistic Inquiry, 2. 117-124. Carden, Guy, 1970a. "0n Post-Determiner Quantifiers," Linguistic Inquiry, 1. 415—427. , 1970b. "A Problem with Primacy," Linguistic Inguiry, 1. 527-533. Chafe, Wallace L., 1970. Meaning and the Structure gI Language. Chicago: University of Chicago Press. Chomsky, Noam, 1957. Syntactic Structures. The Hague: Mouton and Company. , 1959. "Review of Skinner's Verbal Behavior," Lanaaasg. 35. 26-58. , 1964. Current Issues Ig Linguistic Theory. The Hague: Mouton and Company. , 1965. Aspects pi the Theory g: Syntax. Cambridge, Mass.: M.I.T. Press. , 1966. Topics Ig the Theory 9; Generative Grammar. The Hague: Mouton and Company. , 1968. Language and Mind. New York: Harcourt, Brace and World, Inc. , 1970a. "Remarks on Nominalization." In Jacobs and Rosenbaum. 184-221. , 1970b. "Deep Structure, Surface Structure and Semantic Interpretation." Roman Jakobson and Shigeo Kawamoto (eds.), Studies Ig General and Oriental Linguistics: Presented pg Shiro Hattori gp the Occasion pf His Sixtieth Birthday. Tokyo: TEC Company. 52-91. , 1971. "Conditions on Transformations." Bloomington: Indiana University Linguistics Club. , 1972. "Some Empirical Issues in the Theory of Transformational Grammar." Noam Chomsky, Studies gg Semantics Ig Generative Grammar. The Hague: Mouton and Company. 120-202. , and Morris Halle, 1968. The Sound Pattern gI English. New York: Harper & Row, Publishers. 388 Emonds, Joseph E., 1970. Root and Structure-Preserving Transformations. Unpublished doctoral dissertation, IVIOIOT. Fillmore, Charles J., 1965. "Entailment Rules in a Semantic Theory," Project pp Linguistic Analysis Report IQ. Columbus: Ohio State University. 60-82. , 1966. "Deictic Categories in the Semantics of 'come'," Foundations gf Langpage, 2. 219-227. , 1968. "The Case for Case." In Bach and Harms. 1-88. Fodor, Jerry A., 1965. "The Structure of a Semantic Theory," Lapguage, 59. 170-210. , and Jerrold J. Katz (eds.), 1964. The Structure g: Language: Readings in the Philosophy 9: Lagguage. Englewood Cliffs, N.J.: Prentice-Hall, Inc. Gleitman, Lila, 1965. "Coordinating Conjunctions in English," Lan ua e, 51. 260-293. Green, Georgia, 1968. "0n.Igg and Either, and Not Just Igg and Either, Either," Papers from the Fourth Regional Meeting Chicago Linguistic Society. Chicago: University of Chicago. 22-59. Greenberg, Joseph H., 1966. fLanguage Universals." Thomas A. Sebeok (ed.), Current Trends Ig Linguistics, Vol. III: Theoretical Foundations. The Hague: Mouton and Company. 61-112. Gruber, Jeffrey, 1967. "Topicalization in Child Language," Foundations g: Lan a e, 5. 57-65. Harris, Zellig, 1957. "Co-Occurrence and Transformation in Linguistic Structure," Lan a e, 33. 293-340. Hasegawa, Kinsuke, 1968. "The Passive Construction in English," Langgage, 44. 250-245. , 1972. "Transformations and Semantic Interpreta- tion," Linguistic Inquiry, 3. 141-159. Inoue, Kazuko, 1969. I Study pf Japanese Syntax. The Hague: Mouton and Company. Jackendoff, Ray, 1967. "An Interpretive Theory of Pronouns and Reflexives." Mimeographed, M.I.T. 389 Jackendoff, Ray, 1969. "An Interpretive Theory of Negation," Foundations gI Lan.ua e, 5. 218-241. , 1971. "On Some Questionable Arguments about Quantifiers and Negation," Langgage, 47. 284-297. Jacobs, Roderick A., and Peter S. Rosenbaum, 1968. English Transformational Grammar. Waltham, Mass.: Blaisdell Publishing Company. , 1970. Readings Ig English Transformational Grammar. Waltham, Mass.: Ginn and Company. Kajita, Masaru, 1968. g Generative-Transformational Study g: Semi-Auxiliaries Ig Present-Day American English. Tokyo: Sanseido Company Ltd. Katz, Jerrold J., 1965. "Recent Issues in Semantic Theory," Foundations gI Language, 3. 124-194. , and Paul M. Postal, 1964. gg Integrated Theory 9: Linguistic Descriptions. Cambridge, Mass.: M.I.T. Press. King, Harold V., 1970. "On Blocking the Rules for Contraction in English," Linguistic Inquiry, 1. 154-136. Kiparsky, Paul, 1968. "Tense and Mood in Indo-European Syntax," Foundations g: Lan/ua e, 4. 50-57. , and Carol Kiparsky, 1970. "Fact." Manfred Bierwisch and Karl E. Heidolph (eds.), Recent Developments in Linguistics. The Hague: Mouton and Company. 145-175. Klima, Edward, 1964. "Negation in English." In Fodor and Katz. 246-523. Kuno, Susumu, 1970. Notes on Japanese Grammar (Part I). Report No. NSF 27 to the National Science Foundation. The Computation Laboratory of Harvard University. Cambridge, Mass. , 1971. "The Position of Locatives in Existential Sentences," Linguistic Inquiry, 2. 533-378. Kuroda, S. Y., 1965. "Causative Forms in Japanese," Foundations g: Lapguage, 1. 50-50. , 1969a. "English Relativization and Certain Related Problems." In Reibel and Schane. 264-287. Kuroda, Lakoff, Lakoff, 390 S. Y., 1969b. ”Attachment Transformations." In Reibel and Schane. 331-351. , 1970. "Some Remakrs on English Manner Adverbials." Roman Jakobson and Shigeo Kawamoto (eds.), Studies Ig General and Oriental Linguistics: Presented pg Shiro Hattori gg the Occasion gI HIg Sixtieth Birthday. Tokyo: TEC Company. 378-596. George, 1965. Qg the Nature g: Syntactic Irregular- i y. Report No. NSF 16 to the National Science Foundation. The Computation Laboratory of Harvard University. Cambridge, Mass. , 1968. "Instrumental Adverbs and the Concept of Deep Structure," Foundations g: Lagguage, 4. 4-29. , 1969. "On Derivational Constraints," Papers from the Fifth Regional Meeting Chicagg Linguistic Society. Chicago: University of Chicago. 117-159. , 1970a. "Pronominalization, Negation and the Analysis of Adverbs." In Jacobs and Rosenbaum. , 1970b. "Repartee or a Reply to 'Negation, Conjunction and Quantifiers'," Foundations g: Language, 6. 389-422. , 1971a. "On Generative Semantics." Leon A. Jakobovits and Danny D. Steinberg (eds.), Semantics: gg Interdisciplinary Reader Ig Philosophy, Linguistics and Psychology. Cambridge: Cambridge University Press. 252-296. , 1971b. "On Global Nature of Nuclear Stress Rule." Mimeographed. , and John R. Ross, 1967. "Is Deep Structure Necessary?" Mimeographed, M.I.T. , and David Gordon, 1971. "Conversational Postu- 1ates," ngers from the Seventh Regional Meeting Chicago Linguistic Society. Chicago: Chicago Linguistic Society. 65-84. Robin, 1969a. "A Syntactic Argument for Negative Transportation," ngers from the Fifth Regional Meeting Chicago Linguistic Society. Chicago: University of Chicago. 140-147. , 1969b. "Some Reasons Why There Can't Be Any some-any Rule," Langgage, 45. 608-615. 391 Lakoff, Robin, 1970. "Tense and Its Relation to Partici- pants," Langua e, 46. 858-849. Langacker, Ronald, 1969. "On Pronominalization and the Chain of Command." In Reibel and Schane. 160-186. , 1972. Fundamentals g: Linguistic Analysis. New York: Harcourt Brace Jovanovich, Inc. Langendoen, D. Terence, 1969. The Study g: Syntax. New York: Holt, Rinehart and Winston, Inc. Lindholm, James M., 1969. "Negative Raising and Sentence Pronominalization," Papers from the Fifth Regional Meeting Chicago Linguistic Society. Chicago: University of Chicago. 148-158. McCawley, James D., 1967. "Meaning and the Description of Language," Kotoba pg Uchu, 2. 9-11. , 1968a. "Concerning the Base Component of a Transformational Grammar," Foundations gp Language, 4. 243-169. , 1968b. "Lexical Insertion in a Transformational Grammar without Deep Structure," Papers from the Fourth Regional Meeting Chicagp Linguistic Society. Chicago: University of Chicago. 71-80. , 1968c. "The Role of Semantics in a Grammar." In Bach and Harms. 124-169. , 1970. "Where Do Noun Phrases Come From?" In Jacobs and Rosenbaum. 166-185. Martin, Samuel B., 1952. Morph0phonemics g: Standard Colloquial Japanese, Language Dissertation, No. 47. Baltimore. Mikami, Akira, 1960. pg pg Hana gg Nagai. Tokyo: Kuroshio Shuppan. Miller, Roy A., 1972. "Review of Inoue's A Study gf Jgpanese Syntax," Language, 48. 214-230. Partee, Barbara H., 1970. "Negation, Conjunction and Quantifiers: Syntax vs. Semantics," Foundations g: La ua e, 6. 155-165. Perlmutter, David, 1971. Deep and Surface Structure Constraints Ig Syntax. New York: Holt, Rinehart and Winston, Inc. 392 Perlmutter, David, and John R. Ross, 1970. "Relative Clauses with Split Antecedents," Linguistic Inquiry, la 127-1280 Pope, Emily, 1971. "Answers to Yes-No Questions," Linguistic Inquiry, 1. 69-82. Postal, Paul M., 1969. "On So-Called 'Pronouns' in English." In Reibel and Schane. 201-224. , 1970. "On the Surface Verb 'Remind'," Linguistic Reibel, David A., and Sanford A. Schane (eds.), 1969. Modern Studies Ig English. Englewood Cliffs, N.J.: Prentice-Hall, Inc. Rosenbaum, Peter S., 1967. The Grammar gp English Predicate Complement Constructions. Cambridge, Mass.: M.I.T. Press. Ross, John R., 1967a. Constraints gg Variables Ig Syntax. Unpublished doctoral dissertation, M.I.T. , 1969a. "0n the Cyclic Nature of Pronominaliza- tion." In Reibel and Schane. 187-200. , 1969b. "A Proposed Rule of Tree—Pruning." In Reibel and Schane. 288-299. , 1970. "On Declarative Sentences." In Jacobs and Rosenbaum. 222-272. Searle, John R., 1969. Speech Acts: gg Essay Ig the Philosophy gp Language. Cambridge: Cambridge University Press. Soga, latsuo, 1966. Some Syntactic Rules gp Modern Colloquial Japanese. Unpublished doctoral disser- tation, Indiana University. Tokieda, Motoki, 1941. Kokugogaku Genron. Tokyo: Iwanami Shoten. Weinreich, Uriel, 1966. "Explorations in Semantic Theory." Thomas A. Sebeok (ed.), Current Trends Ig Linguistics, Vol. III: Theoretical Foundations. The Hague: Mouton and Company. 395-477. 95 I|I “ H I“ u “I I" “ AII I“ NIIIIII A“ AM“ HI 3 1293 0306