The University of Texas at Austin; College of Liberal Arts
Hans C. Boas, Director :: PCL 5.556, 1 University Station S5490 :: Austin, TX 78712 :: 512-471-4566
LRC Links: Home | About | Books Online | EIEOL | IE Doc. Center | IE Lexicon | IE Maps | IE Texts | Pub. Indices | SiteMap

Syntactic Typology: Studies in
the Phenomenology of Language

Winfred P. Lehmann


2. Japanese:
A Characteristic OV Language

Susumu Kuno

2.0. Preface

This paper gives a typological sketch of the syntactic structure of modern standard Japanese. Discussion of each typological feature is necessarily brief and superficial. There are many important features of Japanese that this paper gives only passing mention, or excludes from discussion. An extensive and in-depth description of the structure of the language can be found in Martin 1975. Treatments of selected features of Japanese in the framework of generative theory of grammar can be found, for example, in Kuroda 1965 and 1972b, Kuno 1973, and Shibatani, ed., 1976a. Two periodicals, Papers in Japanese Linguistics (Department of Linguistics, University of Southern California), and Journal of the Association of Teachers of Japanese (Department of Far Eastern Languages and Civilizations, University of Chicago), regularly carry papers in Japanese linguistics.

2.1. Structure of Simple Clauses

2.1.1. Basic Word Order

Japanese is a verb-final language. Word order in the sentence is relatively free, as long as the sentence ends with a main verb (see section 2.1.2 for nonverb-final sentences involving afterthoughts). For example, 'John introduced Tom to Mary' in Japanese can have the following word-order variations:

(1) a. John ga Mary ni Tom o syookaisita.
  b. John ga Tom o Mary ni syookaisita.
  c. Mary ni John ga Tom o syookaisita.
  d. ?Mary ni Tom o John ga syookaisita.
  e. Tom o John ga Mary ni syookaisita.
  f. ?Tom o Mary ni John ga syookaisita.
Ga, o, and ni are postpositional particles representing the nominative, accusative, and dative case marking. The above sentences are identical in logical content, but are different in discourse presupposition in a very subtle way. Ordinarily, constituents that represent older information precede those that represent newer information.

The subject-initial sentence pattern is the most common among the various word order patterns. In a large-scale sentence-pattern count of modern Japanese journalistic writings it was found that sentences with SOV word order occur seventeen times more frequently than sentences with OSV order. (This is a cross-language characteristic, as observed in Greenberg 1963, Greenberg's Language Universal 1: In declarative sentences with nominal subject and object, the dominant order is almost always one in which the subject precedes the object. Greenberg's observation is not without exception. For example, Keenan 1976a establishes Malagasy as having a basic VOS order. Derbyshire 1977 shows that Hixkaryana, a Carib language spoken in northern Brazil, has a basic order of OVS.) Despite the existence of statistical facts such as this, it is in general extremely difficult to establish an underlying word order of constituents of a free word order language like Japanese. There is, however, at least one syntactic argument for hypothesizing SOV, and not, say, OSV, as representing the underlying word order for Japanese; see Kuno 1971 for underlying word order of SOV sentences and existential sentences. Japanese has a small class of verbals, all stative, that mark their object, as well as their subject, with the particle ga.

(2) a. John ga Mary ga suki na koto
fond-of is fact-that
  'the fact that John is fond of Mary'

A sentence with a stative predicate whose subject is marked with ga instead of the thematic particle wa acquires the interpretation of exhaustive listing, i.e., 'x and only x'. Thus,

(2) b. John ga Mary ga suki da.
  fond-of is
means '(Among those under discussion,) John and only John likes Mary; it is John who likes Mary.' "Subject NP + ga" with stative predicates in subordinate clauses does not have to receive this exhaustive listing interpretation. Example (2a) is given in the complement clause form to avoid this interpretation. See Kuno 1973, Chapter 2.

The fact that (2a) can mean only that John is fond of Mary, and not that Mary is fond of John, shows that the subject and the object cannot switch word order when they are marked with identical case markers. If we assume that the underlying word order is SOV, this is a perfectly natural constraint attributable to the "anti-ambiguity" factor. On the other hand, had we assumed that the underlying word order was OSV, we would need an obligatory "word-order-switch" requirement for instances where the subject and the object are marked with the same case marker. There is no natural explanation for the existence of such a constraint.

2.1.2. Word Order in Colloquial Speech

In written Japanese, sentences are almost exclusively SOV, but in colloquial speech nonverbal elements can appear after the main clause verb. For example, observe the following sentences:

(3) a.
Kimi (wa) kono hon (o) yonda ↗.
you (Theme) this book (Acc.) read
'Have you read this book?'
  b.
Kimi (wa) yonda ↗ kono hon (o) ↘.
you   read this book  
  c.
Yonda kimi (wa) kono hon (o) ↘.
read you   this book  
In colloquial speech, the particles wa (thematic), o (accusative) and ga (when used to mark the object of stative verbals) are often deleted; see Kuno 1972 for conditions on particle deletion. What is noteworthy about (3b,c) is that the rise in intonation characteristic of an interrogative sentence is placed on the verb, and not on the last word of the sentence.

Postverbal constituents can be elements in subordinate clauses, as can be seen in the following examples:

(4) a.
Kimi (wa) [kono-aida ano resutoran de nani (o)
you other-day that restaurant at what
tabeta ka] oboete iru ↗.
ate Q remembering are
'Do you remember what we ate at that restaurant the other day?'
  b.
Kimi (wa) [nani (o) tabeta ka] oboete iru
you   what   ate Q remembering are
kono-aida ano resutoran de ↘.
other-day that restaurant at
  c.
[Nani (o) tabeta ka] oboete iru kimi
what   ate Q remembering are you
kono-aida ano resutoran de ↘.
other-day that restaurant at
Postverbal kono-aida 'the other day' and ano resutoran de 'at that restaurant' in (4b,c) are constituents which appear in the interrogative clause in (4a). If these sentences were to be derived by a movement transformation (i.e., by Right Dislocation), Japanese would constitute a serious counterexample to Ross's (1967) hypothesis that all rightward movements are upward bounded.

For example, observe the following sentences:

(i) Complex NP Shift:
  a. [That they elected the man who was absolutely incompetent the president of the company] was obvious.
  b. [That they elected president the man who was absolutely incompetent] was obvious.
  c. *[That they elected president] was obvious the man who was absolutely incompetent.
(ii) Extraposition of Prepositional Phrases:
  a. [That a review of the book appeared] was surprising.
  b. [That a review appeared of the book] was surprising.
  c. *[That a review appeared] was surprising of the book.
An exception to Ross's principle has been found in Navajo (Kaufman 1974), but only with respect to movement of special grammatical formatives. As far as I know, no language has been found that can move regular lexical items rightward in violation of Ross's principle.

There is strong indication, however, that (3b,c) and (4b,c) are generated not by Right Dislocation, but by a process that adds afterthoughts to the end of a sentence. Namely, in (3b,c), for example, the speaker first assumes that the hearer can understand what is meant by

(5) a.
Kimi yonda ↗.
you read
'Did you read (it)?'
  b.
Yonda ↗.
read
'Did (you) read (it)?'
The speaker adds kono hon 'this book' and kimi kono hon 'you this book' to (5b) and (5c), respectively, to make sure that the hearer will correctly interpret the reference of missing subject and object. Similarly, in (4b), for example, the speaker first utters:
(6) Kimi [nani (o) tabeta ka] oboete iru ↗.
  you what   ate Q remembering are
  'Do you remember what we ate?'
assuming that the hearer will understand the question as referring to 'the other day' and 'at that restaurant', later adding the two postverbal elements to make sure that there will be no misunderstanding.

The "afterthought" analysis of nonverb-final sentences makes it possible to make the following two predictions:

(i) Postverbal elements are either discourse-predictable (or rather, the speaker assumes that they are) or supplementary; therefore, the sentences should have made sense without them.
(ii) Elements that would change the interpretation of the first part of the sentence cannot appear postverbally.
These predictions are borne out by the following examples:
(7) a.
Kimi nani taberu ↗.
you what eat
'What are you going to eat?'
  b.
*Kimi taberunani.
(8) a.
Boku Nihon ni sando sika itta koto
I Japan to thrice only went experience
ga nai.
  have-not
'I have been to Japan only three times.'
  b.
*Boku Nihon ni itta koto ga nai,
I Japan to went experience   have-not
sando sika.
thrice only
(7b) shows that wh-words cannot appear postverbally. This is because (i) wh-words are discourse-nonanaphoric; and (ii) the postverbal addition of a wh-word would change the interpretation of the first part of the sentence completely, from that of a yes-or-no question (i.e., 'Did you eat?') to that of an interrogative-word question (i.e., 'What did you eat?'). (8a) shows that sika 'only' requires that a negative follow it; thus, the more accurate translation of sika would be 'any more than'. (8b) is ungrammatical because the first part of the sentence (up to the verb) states that the speaker has never been to Japan, while the subsequent addition of sando sika 'only three times' forces a complete switch in interpretation.

Postverbal elements never appear in subordinate clauses, even in colloquial speech; (9b) is totally ungrammatical.

(9) a.
Kimi [Taroo ga Hanako to kekkonsita] koto
you   with married that
sitte iru ↗.
knowing are
'Do you know that Taroo married Hanako?'
  b.
*Kimi [Taroo ga kekkonsita Hanako to ] koto
you   married   with   that
sitte iru ↗.
knowing are
  c.
Kimi [Taroo ga kekkonsita] koto sitte iru
you   married that knowing are
Hanako to ↘.
  with

2.1.3. Simple Sentence Types

Japanese sentences can be divided into three categories depending upon whether their verbals are (i) verbs, (ii) adjectives, or (iii) copulas. All three of these types of verbals conjugate with respect to tense, mode, and subordinating types:

(10) Verb   Adjective Copula
  'eat' 'read' 'be young' 'be'
Nonpast tabe-ru yom-u waka-i da
Past tabe-ta yon-da wakak-at-ta dat-ta
Suppositional/Intentional tabe-yoo yom-oo wakak-ar-oo dar-oo
Imperative tabe-ro,-yo yom-e
Subjunctive tabe-reba yom-eba wakak-e-reba nara
Gerundive tabe-0 yom-i wakak-u de, ni
Continuative tabe-te yon-de wakak-u-te
Examples of the three types of sentences appear in the following sections.

2.1.3.1. Verbs

(11) a. S V:
Taroo ga kita.
  came
'Taroo came.'
  b. S O V:
Taroo ga tegami o kaita.
  letter   wrote
'Taroo wrote a letter.'
  c. S IO DO V:
Taroo ga Hanako ni tegami o kaita.
  to letter   wrote
'Taroo wrote a letter to Hanako.'
  d. S IO V:
Taroo ga Hanako ni atta.
  to met
'Taroo met Hanako.'
Transitive constructions in Japanese acquire animate subjects. For example,
(12) Taihuu ga ie no hei o kowasita.
  typhoon   house 's fence   destroyed
  'The typhoon destroyed the house's fence.'
is an extremely unnatural sentence with a distinct flavor of direct translation from English. Instead of constructions like this, intransitive inchoative constructions are normally used in colloquial speech:
(13) Taihuu de ie no hei ga kowareta.
  typhoon with house 's fence   broke
  'Because of the typhoon, the house's fence (Lit.) broke.'
There are, however, some transitive constructions with inanimate subjects that have become part of the language, especially in writing. Most of these constructions have abstract nouns as their subjects:
(14) a.
Sensoo ga ooku no hito no unmei o kaeta.
war   many   people 's fate   changed
'The war changed many people's fates.'
  b.
Koi ga Taroo o moomoku ni sita.
love   blind being did
'Love made Taroo blind.'

2.1.3.2. Adjectives

Adjectives are used as main clause verbals and are not followed by copulas:

(15) a.
Taroo wa mada wakai.
  still is-young
'Taroo is still young.'
  b.
Kotosi no huyu wa samui.
this-year 's winter   is-cold
'It is cold this winter.'
There are a few transitive adjectives. Their objects are marked with the particle ga, which, as we have already seen, is ordinarily used for marking the subject of a sentence.
(16) a.
Taroo ga Hanako ga kirai na koto
  hateful is fact-that
'the fact that Taroo dislikes Hanako'
  b.
boku ga okane ga hosii koto
I   money   am-desirous fact-that
'the fact that I want money'

2.1.3.3. Copulas

(17) a.
Taroo wa sensei da.
  teacher is
'Taroo is a teacher.'
  b.
Taroo wa sensei datta.
  teacher was
'Taroo was a teacher.'
Japanese also has a word class called nominal-adjective. Nominal-adjectives are adjectival in meaning, but do not conjugate, and are followed by copulas:
(18) a.
Kono toori wa sizuka da.
this street   quiet is
'This street is quiet.'
  b.
Taroo wa syooziki da.
  honest is
'Taroo is honest.'

The present tense copula da (but not its "polite" counterpart desu) is deleted before certain sentence-final particles; see section 2.1.4 for desu. For example, observe the following sentences:

(19) Ka: sentence-final particle for questions
  a.
Taroo wa kuru ka.
  come Q
'Is Taroo coming?'
  b.
Taroo wa syooziki datta ka.
  honest was Q
'Was Taroo honest?'
  c.
(20) Sa: sentence-final particle meaning 'I assure you'
  a.
Taroo wa kitto kuru sa.
  certainly come  
'I assure you that Taroo will come without fail.'
  b.
Taroo wa tensai daroo sa.
  genius will-be  
'I assure you that Taroo will (must) be a genius.'
  c.
Sentences such as (19c) and (20c) have the semblance of being sentences without verbals, but the absence of verbals in these surface sentences is simply due to a low-level copula deletion rule, as can be seen by comparing them with the corresponding past tense and suppositional sentences.

Another characteristic of copulative sentences is that precopular constituents cannot be moved around:

(21) a.
Tanaka ga sono hon o katta.
  that book   bought
'Tanaka bought that book.'
  b.
Sono hon o Tanaka ga katta.
that book   bought
(22) a.
Tanaka wa tensai da.
  genius is
'Tanaka is a genius.'
  b.
*Tensai Tanaka wa da.
genius   is
Similarly, no elements can be inserted between the copula and the precopular constituents:
(23) a.
Tanaka wa sono hon (o) mo katta.
  that book   also bought
'Tanaka bought that book, too.'
  b.
*Tanaka wa tensai mo da.
  genius also is
'Tanaka is a genius, too.'

2.1.4. Agreement

Verbs, adjectives, and copulas do not show number, person, and gender agreements. They do show, however, agreement with respect to levels of honorificness and politeness. Respect for the referent of the subject is typically expressed by prefixing the gerundive form of a verb with o- (honorific prefix) and adding ni nar(-u) (Lit., 'become being') to the right of the form, and by prefixing the adjectival stem with o-.

(24) a.
Taroo ga sono tegami o yon-da.
  that letter   read-Past
'Taroo read the letter.'
  b.
Yamada- sensei ga sono tegami o o- yom- i ni nat- ta
  teacher   that letter   read   Past
'Teacher Yamada read the letter.' (Respect for Teacher Yamada)
(25) a.
Taroo wa waka-i.
  be-young-Nonpast
'Taroo is young.'
  b.
Yamada- sensei wa o-waka-i
  teacher   be-young-Nonpast
'Teacher Yamada is young.' (Respect for Teacher Yamada)
Respect for the referent of a nonsubject noun phrase (usually the object or dative) is expressed by prefixing the gerundive form of a verb with o- and adding su(-ru) 'do' to the right of it.
(26) a.
Taroo ga Hanako ni sono tegami o yon-da.
  to that letter   read-Past
'Taroo read the letter to Hanako.'
  b.
Taroo ga Yamada- sensei ni sono tegami o o- yom-
  teacher to that letter   read
i si- ta.
  Past
'Taroo read the letter to Teacher Yamada.' (Respect for Teacher Yamada)

Respect for the hearer is expressed by using the polite form of verbals, which are formed by adding mas(-u) to the right of the gerundive form of a verb, by using the suppletive form des(-u) for the copula, and by adding des(-u) to the right of the nonpast or past form of an adjective:

(27) a.
Taroo ga sono tegami o yom-u. (Informal level)
  that letter   read-Nonpast
'Taroo reads the letter.'
  b.
Taroo ga sono tegami o yom-i mas-u. (Polite to the hearer)
(28) a.
Taroo wa tensai da. (Informal level)
  genius is
'Taroo is a genius.'
  b.
Taroo wa tensai des-u. (Polite to the hearer)
(29) a.
Taroo wa mada waka-i. (Informal level)
  still be-young-Nonpast
'Taroo is still young.'
  b.
Taroo wa mada waka-i des-u. (Polite to the hearer)

2.1.5. Subjectless Sentences

Japanese allows deletion of subjects that are recoverable from linguistic or nonlinguistic context. For example, observe the following sentences:

(30) a.
Soo omoimasu.
so think
'I think so.'
  b.
Kono hon o katte ageyoo.
this book   buying give-will
'I will buy you this book.'
  c.
Amerika ni ikitai.
  to go-want
'I want to go to America.'
  d.
Amerika ni ikitai ka.
  to go-want Q
'Do you want to go to America?'
  e.
Taroo no ie ni ittara, rusu datta.
  's house to went-when absent was
'When I went to Taroo's house, he was not home.'
The first and second person pronouns are rather freely deleted because they are readily recoverable from discourse context. (30e) allows the deletion of a third person subject, since its antecedent is in the same sentence.

Japanese has at least two types of subjectless sentences that cannot be attributed to deletion of recoverable subjects. The first type includes sentences of the following kind:

(31) a.
Oya, ame da.
oh rain is
'Oh, (Lit.) is rain.' 'Oh, it is raining.'
Cf. Oya, ame ga hutte iru.
  oh rain   falling is
  'Oh, it is raining.'
  b.
Doroboo da!
thief is
(Lit.) 'Is a robber!' 'Robber!'
The x da is the template that Japanese uses in giving minimal information while retaining the sentencehood of the statement. The same pattern is used in answering wh-questions:
(32) a.
Kimi wa kinoo doko e itta ka.
you   yesterday where to went Q
'Where did you go yesterday?'
  b.
Boston da
(Lit.) 'Is Boston.'
(33) a.
Dare kara tegami ga kita ka.
who from letter   came Q
'From whom did a letter come?'
  b.
Taroo (kara) da.
(Lit.) 'Is (from) Taroo.'

The second type of subjectless sentences can be exemplified by the following sentence:

(34)
Boku ni wa, subete ga owatta yoo ni
I to   all   ended appearance being
 
omow-are-ru.
think-Passive-Nonpast
 
(Lit., 'To me, is thought in the appearance of everything having come to an end.') 'It seems to me that everything has come to an end.'
Boku 'I' is marked with a dative particle representing the experiencer, and therefore cannot function as a surface subject. Subete ga owatta 'everything has come to an end' is an adjectival clause that modifies yoo 'appearance', a formal (grammatical) noun, which is followed by the adverbial form ni of the copula. Therefore, ...yoo ni cannot function as a surface subject either. Thus, the sentence is completely devoid of a surface subject.

Some super-honorific sentences also lack surface subjects:

(35) a.
Tennoo-heika wa kaze o o-hik-i
emperor cold   Honorific-catch-ing
asobas-are-ta.
do-Honorific-Past
'The Emperor has caught a cold.'
  b.
Tennoo-heika ni wa kaze o o-hik-i
emperor to   cold   Honorific-catch-ing
asobas-are-ta.
do-Honorific-Past
(Lit.) 'To the Emperor, has caught a cold.'
The super-honorific form of a verb is made by prefixing o- to the gerundive form and adding asobas(-u) to the right of it. (35a) has tennoo-heika 'emperor' as its subject, but (35b) is subjectless. The sentence pattern of (35b) is due, it seems, to a desire to avoid holding a person of high honor responsible for an action as an agent.

2.1.6. Sentences with and without Themes

The theme of a sentence is marked with the postpositional particle wa. Observe the following two sentences:

(36) a. John ga kita.
'John came.'
  b. John wa kita.
(36a) is a sentence which does not have a theme. That is, it is not a statement about John. Rather, it is a statement that presents the whole event described as something new, or alternatively, it presents John as the person who fulfills the template "it is x that came." On the other hand, (36b) has John as the theme of the sentence. It is a statement about John.

There are sentences with themes that do not have corresponding themeless sentences. For example, observe the following. (37b) is due to Mikami 1970.

(37) a.
Sakana wa tai ga ii.
fish   redsnapper   is-good
'Speaking of fish, the redsnapper is the best.'
  b.
Basyo wa okunai-setu ga attooteki datta.
place   indoor-theory   predominant was
'Speaking of the place (of the murder), the indoor-theory was predominant.'
The sentence pattern of (37) is ordinarily called the "double subject" pattern. However, there is no evidence that indicates that Sakana wa and Basyo wa of (37a,b) function as subjects in these sentences. The theme of a sentence which is coreferential with the subject is marked with ga in nominal and adjectival clauses:
(38) a.
Taroo wa wakai.
  is-young
'Taroo is young.'
  b.
Taroo ga wakai.
'It is Taroo who is young.'
  c.
Taroo ga (*wa) wakai koto
'the fact that Taroo is young'
On the other hand, it is impossible to mark sakana 'fish' and basyo 'place' of (37) with ga in embedded clauses:
(39) a.
*sakana ga tai ga ii koto
fish   redsnapper   is-good fact-that
'the fact that (Lit.) speaking of fish, the redsnapper is the best.'
  b.
*basyo ga okunai-setu ga attooteki datta
place   indoor-theory   predominant was
koto
fact-that
'the fact that (Lit.) speaking of the place (of the murder), the indoor-theory was predominant'
This shows that Sakana wa and Basyo wa cannot be regarded as performing double functions of theme and subject. Li and Thompson 1976b include the sentence pattern of (37) as characteristic of topic-prominent languages.

Since (37a,b) do not have themeless source sentences, we have to assume that they are derived from underlying structures that already have themes at the sentence-initial position, as in [Theme [Sentence]].

2.1.7. Double Subject Sentences

Japanese has sentences with double subjects, both marked with the nominative article ga; see Kuno 1973, Ch. 3, for details. Observe the following sentences:

(40) a.
Taroo no otoosan ga sinde simatta.
  's father   dying ended-up
'Taroo's father has died.'
  b.
Taroo ga otoosan ga sinde simatta.
'Taroo — his father has died.'
(41) a.
New York no koogai ni yoi zyuutakuti ga
  's suburbs in good residential-area  
aru.
exist
'In the suburbs of New York, there are good residential areas.'
  b.
New York no koogai ga yoi zyuutakuti ga aru.
(Lit.) 'It is New York's suburbs that there are good residential areas.'
  c.
New York ga koogai ni yoi zyuutakuti ga aru.
(Lit.) 'It is New York that in the suburbs there are good residential areas.'
Since yoi zyuutakuti ga aru 'there are good residential areas' and koogai ni yoi zyuutakuti ga aru 'in the suburbs, there are good residential areas' of (41b,c) are stative predicates, New York no koogai 'New York's suburbs' and New York, with the ga marking, receive the exhaustive listing interpretation of 'It is New York's suburbs that ...; it is New York that ...' (See the discussion of ga marking the subject of stative predicates in 2.1.1.)

It seems that (40b) and (41b,c) have a structure that can be represented as [S [S V]]. That the first subject is not in the same simplex sentence as the verb can be shown by the fact that it does not trigger Simplex Sentence Reflexivization (see Section 2.3.3), and that it does not trigger Honorific Agreement. Observe the following sentences, which are due to Shibatani (1976):

(42) a.
*Yamada- senseii ga musuko ga zibuni ni
  teacher   son   self to
unzarisite iru.
disgusted is
'It is Teacher Yamadai whose son is disgusted with selfi.'
  b.
*Yamada- sensei ga inu ga onakunari ni
  teacher   dog   dying-Honorific  
natte simatta.
becoming ended-up
'It is Teacher Yamada whose dog has died.'
In (42a), the subscript i is used to show that Yamada-sensei and zibun are coreferential. The sentence is ungrammatical in that interpretation. (42b) is ungrammatical because onakunari ni nar-, an honorific form for sin- 'die', refers not to what it is intended for (i.e., Teacher Yamada), but to inu 'dog', the second subject, which does not semantically qualify as recipient of the speaker's respect. These two phenomena can be explained if we hypothesize the [S [S V]] structure for the double subject sentence pattern, and if we assume that Reflexivization and Honorific Marking apply only within the bound of simplex sentences.

Double subject sentences are very different from the "NP-ga + NP-ga + Stative Verbal" sentences that we have seen in Section 2.1.3. While (40b) and (41b,c), as shown in (40a) and (41a), have corresponding single subject sentences, (2), (16a), and (16b) do not have corresponding single subject versions:

(43) a.
Taroo ga Hanako ga suki da.
  fond-of is
'It is Taroo that likes Hanako.'
  b.
*Taroo no/ni Hanako ga suki da.
Similarly, when the first subject is deleted from (40b) and (41b,c), we still obtain complete sentences, as shown below:
(44) a.
Otoosan ga sinde simatta.
father   dying ended-up
'Father has died.'
  b.
Yoi zyuutakuti ga aru.
good residential-area   exist
'There are good residential areas.'
  c.
Koogai ni yoi zyuutakuti ga aru.
suburbs in good residential-areas   exist
'There are good residential areas in the suburbs.'
In contrast, when the first NP-ga is deleted from the "NP-ga + NP-ga + Stative Verbal" pattern, we obtain elliptical sentences:
(45) 0 Hanako ga suki da.
fond-of is
  'I like Hanako; he likes Hanako; etc.'

I give below some more examples of double subject sentences:

(46) a.
Taroo wa atama ga ii.
  head   is-good
'Taroo is bright.'
  b.
Zoo wa hana ga nagai.
elephant   nose   is-long
'Speaking of the elephant, its trunk is long.'
  c.
Taroo wa otoosan ga gakkoo no sensei da.
  father   school 's teacher is
'Speaking of Taroo, his father is a school teacher.'
The interaction of this thematic pattern with the double subject pattern produces sentences which seem peculiar to speakers of languages that have neither of these two patterns:
(47) a.
Hana wa zoo ga nagai.
nose   elephant   is-long
'Speaking of noses/trunks, (Lit.) an elephant is long.'
  b.
Me wa Merii ga aoi.
eye   Mary   is-blue
'Speaking of eyes, (Lit.) Mary is blue.'
At first glance, these sentences appear to have zoo ga 'elephant' and Merii ga as the surface subjects of nagai 'is long' and aoi 'is blue', but in fact, zoo ga and Merii ga are the higher subjects of double subject sentences, in which the lower subjects have been deleted due to coreferentiality with the themes of the sentences:
(48) [Theme [S [S V]]]
  a.
[Hana wa [zoo ga [hana ga nagai ]]]
nose   elephant   nose   is-long  
  b.
[Me wa [Merii ga [me ga aoi ]]]
eye   eye   is-blue  

2.1.8. Adpositions

Japanese, as an SOV language, displays all the characteristics that Greenberg (1963) has attributed to SOV languages. One of the characteristics concerns adpositions. (See Kuno 1974 for perception-oriented explanation for such a universal.)

Greenberg's Language Universal 4: With overwhelmingly greater than chance frequency, languages with normal SOV order are postpositional.
Japanese adpositions are all postpositional. I give below representative samples of adpositions classified according to their functions, excluding those that are used in connecting clauses, which I will discuss in section 2.5.1.
(49) Thematic and Contrastive Particles
  a.
John wa tensai da.
'John is a genius.'
  b.
John wa tensai de wa nai.
  genius being   not
'John is not a genius.'
(50) Quantifierlike Particles
  a.
Taroo mo tensai da.
  also genius is
'Taroo, too, is a genius.'
  b.
Taroo sika konakatta.
  only came-not
'Only Taroo came.'
(51) Noun-Coordinating Particles
  a.
Taroo to Hanako ga kita.
  and   came
'Taroo and Hanako came.'
  b.
Taroo ya Hanako ga kita.
  and   came
'Taroo and Hanako (and others) came.'
  c.
Taroo ka Hanako ga kuru.
  or   come
'Either Taroo or Hanako will come.'
(52) Case-Marking Particles
  a.
Taroo ga zidoosya de Hanako to Tookyoo kara
  Nom. car by   with   from
Hirosima made ryokoosita.
  up-to traveled
'Taroo traveled with Hanako by car from Tokyo to Hiroshima.'
  b.
Taroo no otoosan ga Amerika e itta.
  's father Nom.   to went
'Taroo's father went to America.'
(53) Sentence-Final Particles
  a.
Kimi wa kono hon o yonda ka.
you   this book   read Q
'Did you read this book?'
  b.
Boku wa kono hon o moo yonda yo.
I   this book   already read I-tell-you
'I tell you that I have read this book already.'
  c.
Ano hito wa tensai da naa.
that person   genius is Exclamatory
'Boy, that man is a genius!'
  d.
Kimi wa kinoo gakkoo o yasunda ne.
you   yesterday school   rested Tag Q
'You didn't come to school yesterday, did you?'
Note that Japanese uses the sentence-final particle ka in forming interrogative sentences as seen in (53a). The use of sentence-final question particles is a characteristic of postpositional languages, as observed by Greenberg:
Greenberg's Language Universal 9: With more than chance frequency, when question particles or affixes are specified in position by reference to the sentence as a whole, if initial, such elements are found in prepositional languages, and, if final, in postpositional.
Some particles can occur one after another:
(54) a.
Tookyoo ni wa Taroo ga itta.
  to Theme   went
'To Tokyo, Taroo went.'
  b.
Taroo kara mo tegami ga kita.
  from also letter   came
'A letter came from Taroo, too.'
  c.
Fuzi- san wa koko kara ga itiban yoku mieru.
  Mt.   here from Nom. most well can-see
'Speaking of Mt. Fuji, (Lit.) from here is the best visible; one can see Mt. Fuji best from here.'
  d.
Taroo to Hanako to ni atta.
  and   and to met
'I met Taroo and Hanako.'
  e.
Taroo to Hanako to o syootaisita.
  and   and Acc. invited
'I invited Taroo and Hanako.'
The above uses of Japanese particles shows that they are postpositions rather than suffixes.

The coordinating conjunction to can optionally appear after the last conjunct, as shown in (54d,e). Given A to B to and given "B = C to D to," it is possible to have A to [C to D to] to, as shown in the following:

(55)
[Carter to Mondale] to [Kosygin to Bresnev to ] to
  and   and   and   and   and
 
ga kaidansita.
  conferred
 
'Carter and Mondale as a team and Kosygin and Brezhnev as a team had a meeting.'
(55) has a three-particle sequence to to ga; see Kuno 1973, Ch. 8, for details.

2.1.9. Comparison of Inequality

Japanese adjectives and nominal-adjectives do not inflect with respect to comparative and superlative degrees. Instead, they use optional degree adverbs such as motto 'more', zutto 'far', itiban 'first', and mottomo 'most', which appear to the left of the adjective. The postpositional particle yori 'than' is used as a marker of comparison, with the standard with which the comparison is made preceding yori:

(56) a.
Taroo wa Hanako yori zutto wakai.
than   is-young
  Standard Marker   Adjective
'Taroo is far younger than Hanako.'
  b.
Kono kurasu de wa, Taroo ga mottomo wakai.
this class in   most is-young
'In this class, Taroo is the youngest.'
The order of "Standard — Marker of Comparison — Adjective" exemplified by (56a) agrees with the following observation by Greenberg:
Greenberg's Language Universal 22: If in comparison of superiority, the only order, or one of the alternative orders, is standard-marker-adjective, then the language is postpositional. With overwhelmingly more than chance frequency if the only order is adjective-marker-standard, the language is prepositional.

2.2. Nominal Phrases

Greenberg makes the following observations, among others, with respect to the relative position of adjectival modifiers and head nouns in nominal phrases:

Greenberg's Language Universal 18: When the descriptive adjective precedes the noun, the demonstrative and the numeral, with overwhelmingly more than chance frequency, do likewise.
Greenberg's Language Universal 19: When the general rule is that the descriptive adjective follows, there may be a minority of adjectives which usually precede, but when the general rule is that descriptive adjectives precede, there are no exceptions.
Greenberg's Language Universal 20: When any or all of the items — the demonstrative, numeral, and descriptive adjective — precede the noun, they are always found in that order. If they follow, the order is either the same or its exact opposite.
Greenberg's Language Universal 24: If the relative expression precedes the noun either as the only construction or as an alternative construction, either the language is postpositional or the adjective precedes the noun or both.
Similarly, Lehmann (1973a) makes the following observation:
Lehmann's Structural Principle of Language: Modifiers are placed on the opposite side of a basic syntactic element from its primary concomitant.
According to this principle, nominal modifiers (relative clauses, adjectival and genitive expressions) precede nouns in (S)OV languages and follow them in V(S)O languages because they are placed on the side of the head noun opposite its primary concomitant, namely, V.

All the above observations apply to Japanese. Descriptive adjectives, demonstratives, numerals, and relative clauses all precede their head nouns, without exception. Observe the following examples: The no in these sentences is the adjectival form of the copula, and not a particle; see example (15).

(1) a.
kono ni-satu no hon
this 2-vol. is book
(Lit., 'these books which are 2-volumes')
'these two books'
  b.
*ni-satu no kono hon
2-vol. is this book
(2) a.
kono omosiroi hon
this is-interesting book
(Lit., 'this book which is interesting')
'this interesting book'
  b.
*omosiroi kono hon
is-interesting this book
(3) a.
kono kinoo katta hon
this yesterday bought book
'this book, which I bought yesterday'
  b.
kinoo katta kono hon
yesterday bought this book
'this book, which I bought yesterday'
(3) shows that there is no constraint on the relative order of a relative clause and a demonstrative. It goes without saying that in all the above examples total ungrammaticality results if any of the adjectival expressions appears to the right of the head noun.

Japanese does not have indefinite or definite articles. This does not mean, however, that nouns can be used freely both anaphorically and nonanaphorically. Observe the following sentences:

(4)
Kinoo Sansei-doo de hon o katta. Yuusyoku go
Yesterday bookstore at book   bought dinner after
 
hon o yonda.
book   read
 
'Yesterday, I bought a book at Sanseido Bookstore. After dinner, I read a book/books.'
The second sentence in (4) can mean only 'I read a book; I read books'; it cannot mean 'I read the book.' In order to convey the latter meaning, one has to modify hon 'book' with a demonstrative adjective:
(5) ... Yuusyoku go sono hon o yonda.
dinner after that book   read
  'After dinner, I read that book.'
The above does not mean, however, that Japanese uses a demonstrative adjective whenever English uses the. Observe the following sentences:
(6)
Kinoo se no takai hito to se no
yesterday height   tall person and height Ptc.
 
hikui hito ga tazunete kita. Se no takai hito
low person   visiting came height   tall person
 
wa te ni tue o motte ita.
  hand in stick   having was
 
'Yesterday, a tall man and a short man came to see me. The tall man had a walking stick in his hand.'
In the second sentence, se no takai hito 'the tall man' does not have sono 'that'. In fact, in this context, it is not possible to use any demonstrative adjective. Similarly, observe the following discourse:
(7)
Sono heya ni wa ookii tukue ga atta. Tukue no ue ni
the room in   is-big table   was table 's top in
 
wa, kabin ga atta.
  vase   was
 
'In the room, there was a big table. On the table, there was a vase.'
The second tukue 'table, desk' is not modified by a demonstrative adjective. In contrast to (6), however, it is possible to use sono 'that' in (7). The above examples show that the use of demonstrative adjectives is conditioned by various factors such as whether the noun phrase has its own modifier (adjective or relative clause), whether the noun phrase is used contrastively, or whether it is a locative. The exact conditions for the use of demonstrative adjectives are poorly understood, and await future study.

Japanese nouns do not ordinarily distinguish between singular and plural forms. For example, hon in the first sentence of (4) can mean either 'a book' or 'books'. This does not mean, however, that all nouns can be used freely for both singular and plural. Observe the following discourse:

(8)
Kinoo san-nin no gakusei ga tazunete kita.
yesterday 3-person   student   visiting came
 
 
made hanasikonde itta.
until talking went
 
'Yesterday, three students came to visit me. The students left after having talked with me until four o'clock this morning.'
The "plural" form gakusei-tati 'students' is required in the second sentence, where the noun phrase is anaphoric. The suffix -tati, which is reserved for humans, appears in the following construction also:
(9) Kinoo Taroo-tati ga tazunete kita.
  yesterday visiting came
  'Yesterday, Taroo and others came to visit me.'
It is not well known exactly when plural forms such as gakusei-tati are obligatory, when they are optional, and when they cannot be used.

Japanese has a small number of nouns that form "collective" forms by reduplication: for example, yama-yama 'mountains', ie-ie 'houses', ki-gi 'trees', hito-bito 'people', hana-bana 'flowers'. (A morpheme-initial voiceless consonant becomes voiced in the second half of compounds, b being the voiced counterpart of h.)

2.2.1 Relative Clauses

As I have already mentioned, relative clauses in Japanese always precede their head nouns:

(10) a.
Sono hito ga kono hon o kaita.
that person   this book   wrote
'That person wrote this book.'
  b.
kono hon o kaita hito
this book   wrote person
'the person who wrote this book'
(11) a.
Taroo ga sono hon o yonda.
  that book   read
'Taroo read that book.'
  b.
Taroo ga yonda hon
  read book
'the book that Taroo read'
(12) a.
Taroo ga sono hito to issyoni benkyoosita.
  that person with together studied
'Taroo studied together with that person.'
  b.
Taroo ga issyoni benkyoosita hito
  together studied person
'the person with whom Taroo studied'
The following sentence is ungrammatical:
(13) a.
Taroo ga sono hito to benkyoosita.
  that person with studied
'Taroo studied with that person.'
  b.
*Taroo ga benkyoosita hito
'the person with whom Taroo studied'
(13b) is ungrammatical in the intended interpretation (it is grammatical if it is intended for 'the person that Taroo studied') because the deleted particle to 'with' is not recoverable. In (12b), issyoni 'together' makes it possible to supply this particle. See Kuno 1973, Chs. 20 and 21, for details.
(14) a.
Sono ie no yane wa akai.
that house 's roof   is-red
'The roof of that house is red.'
  b.
yane ga akai ie
roof   is-red house
'the house the roof of which is red'
(15) a.
Sono kodomo wa otoosan ga syoogakkoo no
the child   father   grade-school 's
sensei da.
teacher is
'Speaking of that child, (his) father is a grade-school teacher.'
  b.
otoosan ga syoogakkoo no sensei no kodomo
father   grade-school 's teacher is child
'a child whose father is a grade-school teacher'
In (15b), the no that immediately precedes the head noun kodomo 'child' is not a possessive particle, but is the pre-nominal adjectival form of the copula da. In more formal speech, de aru (Lit., 'is being') appears in its place.

Two facts require special mention in connection with (10) through (15)- First, the particle that marks the noun phrase that is coreferential with the head noun of a relative clause is not present in the surface construction. In other words, Japanese does not allow dangling postpositions. Second, relative clauses in Japanese do not use relative pronouns. Relative clauses (and for that matter, all subordinate clauses) in Japanese are strictly verb-final, and, therefore, verbals signal the end of clauses. There is therefore no need for relative pronouns, whose main function seems to be to mark the clause boundary of embedded clauses. (See Kuno 1974 for details.)

Relativization can enter into relative clauses, adverbial clauses, interrogative clauses, and sentential subjects, namely, into those constructions which normally reject the same process in English:

(16) a.
(Lit.) 'the child who the dog (he) was fond of 0 died'
  b.
(Lit.) 'the book which the publisher who published (it) went bankrupt'
  c.
(Lit.) 'the person who, although (he) died, no one was saddened'
  d.
(Lit.) 'the book which no one knows who wrote (it)'
  e.
(Lit.) 'a person who to see (him) is difficult'
The acceptability of (16a,b) is interesting from a language-typology point of view. Both sentences involve double relativization, namely, relativization from a relative clause. There are many languages which prohibit double relativization, but (16a,b) show that Japanese is not such a language. In (16), head nouns are connected with a link to the corresponding deletion sites in the relative clauses (according to the conventional analysis of relativization). (16a) is said to involve center-embedding double relativization because one link is center-embedded in another, while (16b) is said to involve crisscrossing double relativization because the two links crisscross each other. There are some languages which allow relativization into relative clauses only when the center-embedding pattern of (16a) holds, but the acceptability of (16b) shows that Japanese is not one of these languages. Grammatical strings of the pattern of (16b) abound:
(17) a.
[0 0 sidoosite kurete ita sensei ] ga tenninsite
  advising giving was teacher   transferring
simatta gakusei
ended-up students
(Lit.) 'the students who the teacher who had been teaching (them) moved to another school'
  b.
[0 0 kaita hito ] ga dare da ka wakaranai tegami
  wrote person   who is Q not-know letter
(Lit.) 'the letter such that who the person who wrote (it) is, is not known'

I hypothesize that what is relativized in a Japanese relative clause is not an ordinary noun phrase, but a noun phrase that is the theme of the relative clause. For example, I assume that (18a) below is derived from the intermediate structure represented in (18b) by Theme-Deletion under coreferentiality with the head noun:

(18) a.
Taroo ga yonda hon
  read book
'the book that Taroo read'
  b.
What this analysis assumes is that the head noun must represent the theme of the relative clause; namely, the relative clause must be a statement about the head noun. Justification for this analysis of Japanese relativization is given in Kuno 1973, Ch. 21. Application of the same principle for explaining certain otherwise unexplainable relativization phenomena in English is given in Kuno 1976b.

2.2.2. Genitives

Genitive expressions precede head nouns, as seen in the following:

(19) a.
Taroo no ie
  's house
'Taroo's house'
  b.
Taroo no otoosan no ie
  's father 's house
'Taroo's father's house'
In English, genitive expressions such as the man I met yesterday's wife are rather exceptional (cf. ??the man I met's wife). By contrast, in Japanese, such expressions are very common because the relative clause precedes the head noun, and hence the genitive marker no always follows the head noun of the relative construction.
(20) a.
[ boku ga kinoo atta hito ] no okusan
  I   yesterday met person   's wife
'the man whom I met yesterday's wife'
  b.
[[ boku ga kinoo atta hito ] no okusan ga
  I   yesterday met person   's wife  
tutomete iru kaisya ] no syatyoo
working is company   's president
'the president of the company where the wife of the man that I met yesterday is employed'
  c.
[[[ boku ga kinoo atta hito ] no okusan ga
  I   yesterday met person   's wife  
tutomete iru kaisya ] no syatyoo ga kaita hon ]
working is company   's president   wrote book  
no syuppansya
's publisher
'the publisher of the book that the president of the company where the wife of the man that I met yesterday is employed wrote'
Left branching constructions of the above type are extremely common in Japanese, and they do not cause any difficulty in comprehension.

2.2.3. Numerals and Quantifiers

Numerals can be used as prenominal adjectives, as nouns, and as adverbs. For example, observe the following sentences:

(21) a.
San-nin no yuuzin ga tazunete kita. (Adjectival)
3-person   friend   visiting came  
'Three friends came to visit me.'
  b.
Yuuzin no san-nin ga tazunete kita. (Nominal)
friend 's 3-person   visiting came  
'Three of my friends came to visit me.'
  c.
Yuuzin ga san-nin tazunete kita. (Adverbial)
friend   3-person visiting came  
'Friends came to visit me (Lit.) three-person-ly.'
There are restrictions on the adverbial uses of numerals and quantifiers: they are allowable when they semantically modify noun phrases that are subjects or objects; they are unacceptable when noun phrases of oblique cases are involved. However, the exact conditions for the appearance of adverbial numerals are not well understood:
(22) a.
Yuuzin ga san-nin tazunete kita. (Subject)
friend   3-person visiting came  
'Three friends came to visit me.'
  b.
Yuuzin o san-nin syootaisita. (Object)
friend   3-person invited  
'I invited three friends.'
  c.
*Yuuzin kara san-nin tegami o moratta. (Oblique)
friend from 3-person letter   received  
'I received letters from three friends.'
(23) a.
San-dai no zidoosya de ryookoosita.
3-car   car by traveled
'We traveled in three cars.'
  b.
*Zidoosya de san-dai ryokoosita.
car by 3-car traveled
'We traveled in three cars.'

2.3. Verbal Phrases

2.3.1. Declarative, Interrogative, and Negative

Interrogative sentences are formed by using the sentence-final question particle ka regardless of whether a yes-or-no question or interrogative-word question is involved:

(1) a.
Taroo wa kita ka.
  came Q
'Did Taroo come?'
  b.
Taroo wa sono okane o dare ni yatta ka.
  the money   who to gave Q
'Who did Taroo give the money to?'
  c.
Dare ni Taroo wa sono okane o yatta ka.
who to   the money   gave Q
'Who did Taroo give the money to?'
The interrogative word does not have to be preposed to sentence-initial position. This is a characteristic of SOV languages, as observed by Greenberg:
Greenberg's Language Universal 17: If a language has dominant order VSO in declarative sentences, it always puts interrogative words or phrases first in interrogative-word questions; if it has dominant order SOV in declarative sentences, there is never such an invariant rule.
(1c) seems to have been derived by the same word-order scrambling rule that is responsible for relatively free word order in Japanese as illustrated at the beginning of this paper.

The interrogative word can enter rather freely into coordinate structures, complex noun phrases, adverbial clauses, and sentential subjects. Resulting interrogatives, which are quite natural in Japanese, are almost untranslatable into English.

(2) a.
Taroo to dare to ga kekkonsita ka sitte iru ka?
  and who and   married Q knowing is Q
'Do you know (Lit.) Taroo and who got married?'
  b.
Dare ga dare ni kaita tegami ga itiban
who   who to wrote letter   most
omosirokatta ka.
was-interesting Q
(Lit.) 'The letter which who wrote to whom was most interesting?'
  c.
Taroo wa doko ni itta toki ni kore o katta ka.
  where to went time at this   bought Q
(Lit.) 'At the time that Taroo went where did he buy this?'
Negative sentences are formed by affixing the negative adjectival morpheme (-a-)na to a verbal stem, and by adding the negative adjective na- to the gerundive form of an adjective or copula:
(3) a.
Boku wa kyoo wa nanimo tabe-na-i.
I   today   anything eat-Neg.-Nonpast
'I won't eat anything today.'
  b.
Boku wa kyoo wa gakkoo ni ik-a-na-i.
I   today   school to go-Affix-Neg.-Nonpast
'I don't go to school today.'
(4) a.
Taroo wa wakak-u na-i.
  be-young-ing not-be-Nonpast
'Taroo is not young.'
  b.
Taroo wa tensai de (wa) na-i.
  genius being   not-be-Nonpast
'Taroo is not a genius.'
The negative form of ar- 'exist, have' is na-:
(5) a.
Tukue no ue ni hon ga ar-u.
desk 's top in book   exist-Nonpast
'There is a book on the desk.'
  b.
Tukue no ue ni (wa) nanimo na-i.
desk 's top in   anything not-exist-Nonpast
'There is nothing on the desk.'

2.3.2. Reciprocal Verbs

Japanese has a productive process for forming reciprocal verbs. The verb affix aw- 'match, meet' is added to the gerundive form of action verbs (aw- + -ta > at-ta):

(6) a.
Taroo to Hanako wa tasuke-at-ta.
  and     helping-Recip.-Past
'Taroo and Hanako helped each other.'
  b.
Taroo to Hanako wa hagemas-i-at-ta.
  and encourage-ing-Recip.-Past
'Taroo and Hanako encouraged each other.'
One of the arguments of the coordinate subject of a reciprocal verb can be removed from the subject position and can be made into an adverbial expression:
(7) a.
Taroo wa Hanako to tasuke-at-ta.
  with helping-Recip.-Past
(Lit.) 'Taroo helping-reciprocated with Hanako.'
  b.
Cf. Taroo wa Hanako o tasuke-ta.
Acc. help-Past
  'Taroo helped Hanako.'
  c.
Taroo wa Hanako to hagemas-i-at-ta.
  with encourage-ing-Recip.-Past
(Lit.) 'Taroo encouraging-reciprocated with Hanako.'
  d.
Cf. Taroo wa Hanako o hagemas-i-ta.
Acc. encourage-Affix-Past
  'Taroo encouraged Hanako.'
It is particularly interesting that the noun-coordinating particle to and the comitative particle to are homophonous. See Kuno 1973, Ch. 6 for details.

2.3.3. Reflexive

Japanese has a single reflexive form zibun 'self' for all persons and genders. For the plural, zibun-zati 'selves' is used if the referent is discourse-anaphoric. In simplex sentences, reflexivization is triggered only by the subject of the sentence:

(8) a.
Tarooi ga Hanako o zibuni no ie de korosita.
  self 's house in killed
'Taroo killed Hanako in his (= Taroo's) house.'
  b.
*Taroo ga Hanakoj o zibunj no ie de korosita.
  self 's house in killed
'Taroo killed Hanako in her (= Hanako's) house.'
(9) a.
Hanakoj ga Taroo ni zibunj no ie de korosareta.
  by self 's house in was-killed
'Hanako was killed by Taroo in her (= Hanako's) house.'
  b.
*Hanako ga Tarooi ni zibuni no ie de korosareta.
  by self 's house in was-killed
'Hanako was killed by Taroo in his (= Taroo's) house.'
The fact that (9a) is grammatical while (9b) is not shows that Reflexivization applies after Passivization has applied.

The unambiguity of reflexive inference in (8) and (9) stands in marked contrast with the ambiguity displayed in the following causative sentence. (This was first observed in Akatsuka 1972.)

(10) Hanako ga Taroo ni zibun no ie de benkyoos-ase-ta.
    to self 's house in study-cause-Past
  'Hanako made Taroo study in his/her house.'
This phenomenon can be explained if we assume that a causative sentence involving the causative morpheme (s)ase has a complex underlying structure of the kind informally shown in (11). If the lower sentence has Taroo no ie 'Taroo's house', Reflexivization applies within this clause, with its subject Taroo as trigger. If the lower sentence has Hanako no ie 'Hanako's house', Reflexivization applies after the embedded subject has been marked with ni by Agentive-Ni
(11)
(12)

Attachment, and after the embedded verb has been raised and attached to the main verb sase by Verb Raising, which yields the surface structure shown in (12). At this stage, the second occurrence of Hanako is in the same simplex sentence as the first occurrence of Hanako, and, hence, Reflexivization applies, producing (10), which has zibun as coreferential with Hanako.

The ambiguity of the following "adversity passive" sentence is also in marked contrast with the unambiguity of (9), which is a pure passive sentence:

(13)
Hanako ga Taroo ni zibun no hanasi bakari s-are-ta.
  by self 's talk only do-Passive-Past
 
'Hanako was adversely affected by Taroo's talking about nothing but himself (= Taroo)/herself (= Hanako).'
I assume that pure passive sentences are derived from underlying simplex sentence structures, while adversity passive sentences are derived from complex sentence structures of the kind shown below:
(14)

If we have Taroo no hanasi in the lower clause, Reflexivization applies within the bound of this clause. Next, Agentive-Ni Attachment attaches ni to the lower subject Taroo, and the lower verb is raised and attached to the left of the main clause verb, which triggers tree-pruning processes, yielding a surface structure similar to that of (12). If we have Hanako no hanasi 'Hanako's talk', Reflexivization takes place in the main clause, with the subject Hanako as trigger. See Kuno 1973, Ch. 25, for details.

There are several characteristics of the Japanese reflexive worth mentioning here. First, the reflexive must refer to higher animals. Therefore, Japanese lacks reflexive sentences like History repeats itself. Second, it is difficult to reflexivize a nongenitive NP. Sentences (15) and (16) are ungrammatical.

(15) *Taroo wa zibun o korosita.
self   killed
  'Taroo killed himself.'
Instead, the Sino-Japanese compound zi-satu si-ta 'self-murder did' is used.
(16) *Taroo wa zibun o tataita.
self   hit
  'Taroo hit himself.'
Instead, one has to specify a body part, as in
(17) Taroo wa zibun no atama o tataita.
self 's head   hit
  'Taroo hit his own head.'
Third, reflexive sentences of the pattern of (15) and (16) are passable if the reflexive refers not to a physical object, but to an abstract personality (see Akatsuka 1972 and McCawley 1976 for various characterizations of Japanese reflexives):
(18) a.
Taroo wa zibun o awarenda.
  self   pitied
'Taroo pitied himself.'
  b.
Taroo wa zibun o hihansita.
  self   criticized
'Taroo criticized himself.'
Fourth, Reflexivization in complex sentences is a "point of view" based phenomenon. Observe the following contrast:
(19) a.
Tarooi wa Hanako ga zibuni ni kureta okane o
  self to gave money  
tukatte simatta.
spending ended-up
'Taroo has used up the money that Hanako gave him (= Taroo).'
  b.
*Tarooi wa Hanako ga zibuni ni yatta okane o
  self to gave money  
tukatte simatta.
spending ended-up
'Taroo has used up the money that Hanako gave him (= Taroo).'
The two sentences are differently only with respect to the verbs of the relative clauses: kureta and yatta. They both mean 'gave', but they are different in that the former is a verb used when the speaker describes the action of giving from the point of view of the recipient, while the latter is a verb used when the speaker describes the same action from the point of view of the giver; see Kuno and Kaburaki 1975 and Kuno 1976c. (19b) is unacceptable because there is a conflict in the speaker's point of view: yatta shows that the speaker is describing the event from Hanako's (= the giver's) point of view, while zibun (= Taroo) 'self' shows that the same speaker is describing the event from Taroo's point of view.

2.3.4. Compound Verbs, Adjectives, and Nominal-Adjectives

Japanese is rich in compound verbal formation. I give below representative samples:

(20) Compounding Verbals Added to the Gerundive Form of Verbs
  a. Verb Compounds
yom-i-hazime-ru 'begin to read'
yom-i das-u 'start to read'
yom-i toos-u 'read through'
yom-i a-u 'read to each other'
yom-i oe-ru 'finish reading'
yom-i tuzuke-ru 'continue to read'
yom-i sugi-ru 'read excessively'
  b. Adjective Compounds
yom-i yasu-i 'be easy to read'
yom-i ta-i 'be eager to read'
yom-i zura-i 'be difficult to read'
  c. Nominal Adjective Compounds
yom-i soo da 'look as if ... about to read'
yom-i sugi da 'be excessive in reading'
No elements can be inserted between the gerundive form and the compounding verbals.
(21) Compounding Verbs Added to the Continuative Form of Verbs
  yon-de mi-ru 'try to read'
  yon-de sima-u 'end up reading'
  yon-de yar-u 'read (From the point of view of the agent)'
  yon-de i-ru 'is reading'
  yon-de kure-ru 'read (From the point of view of a nonagent)'
Only certain particles (wa, contrastive, mo 'also', sika 'only', dake 'only', sae 'even') can be inserted between the continuative form and the compounding verbs, as in yon-de mo mi-ru 'try also to read', yon-de sae i-ru 'is even reading'.

The position of compounding verbals relative to the gerundive and continuative forms of verbs illustrated above is consistent with Greenberg's Language Universals 13 and 16:

Greenberg's Language Universal 13: If the nominal object always precedes the verb, then verb forms subordinate to the main verb also precede it.
Greenberg's Language Universal 16: In languages with dominant order VSO, an inflected auxiliary always precedes the main verb. In languages with dominant order SOV, an inflected auxiliary always follows the main verb.
It is also consistent with Lehmann's structural principle given in section 2.2 because it predicts that verbal modifiers for negation, causation, aspectual specifications, etc., are placed after verb roots in (S)OV languages on the opposite side of V from O.

2.3.5. Modality, Aspect, and Tense

The following examples illustrate various ways in which Japanese expresses modality:

(22) Imperative, Necessitative
  a.
Hon o yom-e. (Informal)
book   read-Imper.
'Read books.'
  b.
Hon o yom-i nasai. (Quasi-polite)
book   read-ing do-Imper.
'Read books.'
  c.
Hon o yom-u beki da.
  read-Nonpast should is
'You should read books.'
(Beki is the prenominal attributive form of the Classical Japanese auxiliary verb besi 'should'. When it is used pre-nominally, it is not accompanied by a copula: yomu beki hon 'books to read'.)
(22) d.
Hon o yom-a-nak-e-reba ikenai.
  read-Affix-Neg.-Affix-if is-not-good
(Lit., 'If you ddn't read books, it is not good.')
'You must read books.'
  e.
Hon o yon-de mo yoi.
  read-Cont. even is-good
(Lit., 'Even reading books, it is good.')
'You may read books.'
  f.
Hon o yom-oo.
  read-Intentional
'Let's read books.'
(23) Epistemic Modals
  a.
Ame ga hutta ka mo sir-e-na-i.
rain   fell Q even know-can-Neg.-Nonpast
(Lit., 'One cannot know whether it rained or not.')
'It may have rained.'
  b.
Ame ga huru daroo.
rain   fall I-suppose
'I suppose it will rain.'
(Daroo is the suppositional form of the copula da 'is'.)
  c.
Ame ga huru ni tigai nai.
rain   fall to mistake is-absent
(Lit., 'There is no mistake to the rain falling.')
'It will definitely rain.'
  d.
Ame ga hutta rasii.
rain   fell seem
'It seems to have rained.'
  e.
Ame ga hutta soo da.
rain   fell appearance is
'I hear that it rained.'
As shown in section 2.1.3, verbs, adjectives, and copulas have nonpast and past forms:
(24) a.
Ame ga hur-u.
rain   fall-Nonpast
'It rains; It will rain.'
  b.
Ame ga hut-ta.
rain   fall-Past
'It rained.'
(25) a.
Taroo wa waka-i.
  be-young-Nonpast
'Taroo is young.'
  b.
Taroo wa wakak-at-ta.
  be-young-Affix-Past
'Taroo was young.'
The nonpast tense of an action verb represents either a generic, habitual action or a future action.

There is no tense agreement between the main clause verb and the subordinate clause verb. Observe the following sentences:

(26) a.
Taroo wa siken ga muzukasi-i koto o
  test   be-difficult-Nonpast that  
sit-te i-ru.
knowing be-Nonpast
'Taroo knows that the exam is difficult.'
  b.
Taroo wa siken ga muzukasi-i koto o
  be-difficult-Nonpast  
sit-te i-ta.
  be-Past
  c.
Taroo wa siken ga muzukasik-at-ta koto o
  be-difficult-Past  
sit-te i-ta.
  be-Past
(26b) means 'Taroo knew that the exam was difficult.' That is, the time that the subordinate clause verb refers to is the same as the time that the main clause verb refers to. On the other hand, (26c) means that Taroo knew that the exam had been difficult. That is, the past tense of the subordinate clause refers to a time period prior to the time the main clause verb refers to.

Various periphrastic expressions are used to show aspect. For example, the following sentences show the devices Japanese uses to express the semantics of the present perfective in English:

(27) Completion of Action: e.g., He has just left.
  a.
Taroo ga ki-ta.
'Taroo came, Taroo has come.'
  b.
Taroo ga ki-ta tokoro da.
  come-Past place/moment is
'Taroo has just come.'
  c.
Taroo ga it-te simat-ta.
  go-Cont. end-up-Past
'Taroo has gone.'
(28) Continuation of Action up to Present Time: e.g., I have lived here for a long time.
a.
Boku wa moo zyuunen koko ni sun-de
I   already 10-years have in live-Cont.
iru.
be-Nonpast
'I have lived here for ten years already.'
  b.
Kesa hatizi kara benkyoosi-te i-ru.
this-morning 8 from study-Cont. be-Nonpast
'I have been studying since eight this morning.'
(29) Past Experience: e.g., I have been to Japan once.
  a.
Taroo wa itido Huransu ni it-ta koto ga
  once France to go-Past experience  
ar-u.
have-Nonpast
'Taroo has been to France once.'
  b.
Taroo wa itido mo Huransu ni it-ta koto
  once even France to go-Past experience
ga na-i.
  not-have-Nonpast
'Taroo has never been to France.'

The continuative form followed with i(-ru) 'be' that is used in (28) to express continuation of an action up to the present time is also used to express (i) continuation of action at the present time, (ii) repetition of action, and (iii) succession of the same action:

(30) a.
Hanako ga asoko de nai-te i-ru. (Continuation)
  there at cry-Cont. be-Nonpast
'Hanako is crying there.'
  b.
Hanako wa mainiti tenisu o si-te i-ru. (Repetition)
  every-day tennis   do-Cont. be-Nonpast
'Hanako plays tennis every day.'
  c.
Mainiti oozei no hito ga kaze de sin-de
daily many   people   cold with die-Cont.
i-ru. (Succession)
be-Nonpast
'Many people are dying of colds every day.'

2.3.6. Compound Verbals Representing the Speaker's Attitude

Japanese is a language that often forces the speaker to express an attitude toward the action described in a sentence. For example, it is not possible for a speaker to simply say, 'John visited me.' One must say, 'John came visiting me.' Similarly, it is not possible to say, 'John called me up.' It is necessary to say, (Lit.) 'John came having phoned me,' where came indicates that the action of telephoning was directed toward the speaker. Likewise, it is not easy to say simply, 'John borrowed money from me.' One ordinarily says either 'I lent money to John,' or 'John went away, having borrowed money from me,' where the whole event is seen from the speaker's point of view. The following are Japanese examples illustrating the above point:

(31) a.
*Taroo ga kyoo boku o tazuneta.
  today me   visited
'Taroo visited me today.'
  b.
Taroo ga kyoo boku o tazunete kita.
  today me   visiting came
'Taroo came visiting me today.'
(32) a.
*Taroo ga kyoo boku ni denwa o kaketa.
  today me to phone   called
'Taroo called me up today.'
  b.
Taroo ga kyoo boku ni denwa o kakete kita.
  today me to phone   calling came
(Lit., 'Taroo came having called me up today.')
'Taroo called me up today.'
(33) a.
??Taroo ga kyoo boku kara okane o karita.
  today me from money   borrowed
'Taroo borrowed money from me today.'
  b.
Taroo ga kyoo boku kara okane o karite itta.
  today me from money   borrowing went
(Lit., 'Taroo went having borrowed money from me.')
'Taroo borrowed money from me today.'
The requirement that expressions be speaker-centered is not limited to the use of deictic verbs of coming and going. It is not possible to say neutrally, 'Mary bought me a necktie.' One must say, 'Mary gave me the favor of buying a tie.' Similarly, one seldom says that one will buy a tie for the hearer, but instead says that one will give the hearer the favor of buying a tie, or 'Please give me the favor of buying a tie.'
(34) a.
??Hanako ga boku ni nekutai o katta.
  me to necktie   bought
'Hanako bought me a necktie.'
  b.
Hanako ga boku ni nekutai o katte kureta.
  me to necktie   buying gave
(Lit.) 'Hanako gave me the favor of buying a tie.'
(35) a.
?Kimi ni nekutai o kau.
you to necktie   buy
'I will buy you a tie.'
  b.
Kimi ni nekutai o katte ageru.
you to necktie   buying give
(Lit.) 'I will give you the favor of buying a tie.'
(See Kuno 1973, Ch. 9, for details.)
  c.
Nekutai o kaw-ase-te kudasai.
  buy-caus-ing give-Imper.
(Lit.) 'Please give me the favor of buying a tie (for you).'

Likewise, if John bought chocolates for the speaker's daughter, the speaker does not ordinarily describe this event neutrally, but says, 'John gave me the favor of buying chocolates for my daughter':

(36) a.
??Taroo ga (boku no) musume ni tyokoreeto o
  I 's daughter to chocolate  
katta.
bought
'Taroo bought chocolates for my daughter.'
  b.
Taroo ga (boku no) musume ni tyokoreeto o
  I 's daughter to chocolate  
katte kureta.
buying gave
(Lit.) 'Taroo gave me the favor of buying chocolates for my daughter.'
The above phenomena are all part of a general problem of the speaker's point of view, for which Japanese has an extremely rich lexical and syntactical system. See Kuno 1975, Kuno and Kaburaki 1975, and Kuno 1976b and c.

2.3.7. Passive and Causative

We have already seen some examples of passive and causative sentences. Japanese has two passives: the neutral passive and the adversity passive, both of which have verbs marked with the passive morpheme (r)are(-ru) and an underlying agentive followed by the particle ni.

(37) a.
Taroo ga sensei ni sikar-are-ta. (Pure Passive)
  teacher by scold-Passive-Past
'Taroo was scolded by the teacher.'
  b.
Taroo ga sensei ni musuko o sikar-are-ta.
  teacher by son   scold-Passive-Past
(Adversity Passive)
(Lit.) 'Taroo was adversely affected by the teacher's scolding his son.'
What is interesting here is the fact that the ni that is used to mark the underlying agentive for passive sentences of both kinds is homophonous with the dative marker ni. Observe, further, the following sentence:
(38) a. Underlying Structure
Taroo [saihu (ga) nakunar-] rare-ta.
  purse   disappear Passive-Past
'Taroo was adversely affected by the purse's having disappeared.'
  b. Surface Sentence
*Taroo ga saihu ni nakunar-are-ta.
Although there does not seem to be anything wrong with the underlying structure shown in (38a), the resulting sentence shown in (38b) is totally ungrammatical. This suggests that the ni used both as a dative marker and as a passive agentive marker has the characteristic that it marks a secondary agent. (38b) is ungrammatical because saihu 'purse', which cannot act as an agent, has ended up being followed by this secondary agent marker.

The function of ni as a secondary agent marker can also be seen in the following examples, which involve neither syntactic causatives nor syntactic passives.

(39) a.
Boku wa Hattori- sensei ni gengo-gaku o naratta.
I   teacher   linguistics   learned
'I learned linguistics from Professor Hattori.'
  b.
Boku wa Hattori- sensei ni gengo-gaku
I   teacher   linguistics
o osowatta.
  took-lessons-from
'I learned linguistics from Professor Hattori.'
The ni in these sentences does not represent a goal, direction, or dative of interest; it represents a secondary agent, and as such, alternates with kara 'from'. Similarly, observe the following sentence:
(40) Sono koto o Taroo ni kiita.
  that matter   asked/heard
While the primary interpretation of the sentence is 'I asked Taroo about it,' it can also mean 'I heard about it from Taroo,' although in this interpretation Taroo kara 'from Taroo' is used more often than Taroo ni. The two interpretations of (40) share a common feature, that is, the fact that Taroo is a secondary agent — as the recipient of a question in the case of 'I asked ...', and as the source or originator of message transmission in the case of 'I heard from ...'

Causatives also display very interesting behavior along similar lines. When the underlying embedded clause is transitive, there are two causatives:

(41) a.
Kantoku wa sono siin de haiyuu o nak-ase-ta.
director   that scene at actor Acc. cry-caus-Past
  b.
Kantoku wa sono siin de haiyuu ni nak-ase-ta.
  Dat.  
'The director made the actor cry in that scene.'
In (41a) the underlying embedded clause subject is marked with the accusative particle o, while in (41b) it is marked with the dative particle ni. Shibatani (personal communication, 1976) has observed that (41a) implies that the director was harsh on the actor so that the latter cried, while (41b) implies that the director instructed the actor to cry in that scene. In other words, only the ni-marked causative involves a transmission of a message to induce the action represented by the underlying embedded clause. This observation is consistent with the fact that only the o-marked causative can be used in the following sentences. (I am indebted to S. Tonoike, personal communication, 1976, for this observation.)
(42) a.
Yasai o kusar-ase-te simatta.
vegetables   spoil-cause-Cont. ended-up
(Lit., 'I have had vegetables spoil.')
'Vegetables have been spoiled.'
  b.
Yamada- san wa kazi de kodomo o sin-ase-te
  Mr.   fire by child   die-cause-Cont.
simatta.
ended-up
(Lit., 'Mr. Yamada had his child die because of a fire.') 'Mr. Yamada lost his child in a fire.'
In (42a), there could not have been any message transmission from the speaker to the underlying embedded subject because the latter is inanimate. In (42b), the semantics of the sentence clearly relate that Yamada did not tell his son to die. The use of causatives in both sentences seems to express an underlying feeling on the part of the speaker that the subject of the main clause could have prevented the situation from happening.

I hypothesize that the o-causative and the ni-causative are derived from the underlying structures in (43) (see Kuno 1973, Ch. 27). In (43a), Equi NP Deletion applies and deletes the embedded clause subject, which is identical to the matrix

(43) a. O-Causative ("Hands Off")
  S NP kantoku 'director' NP haiyuu 'actor' S haiyuu 'actor' nak- 'cry' V sase-ta
  b. Ni-Causative ("Message Transmission")
  S S kantoku 'director' S haiyuu 'actor' nak- 'cry' V sase-ta
clause object. Next, Verb Raising applies, which triggers tree-pruning rules to apply, yielding the structure shown below:
(44) S NP kantoku 'director' NP haiyuu 'actor' V V nak- 'cry' V sase-ta 'cause'

Now, the NP for haiyuu 'actor' occupies the object position for the compound causative verb nak-(s)ase-ta 'caused to cry' Hence, it gets marked with the object particle o.

In the case of (43b), Agentive-Ni Attachment applies to the embedded clause subject, followed by Verb Raising. The resulting structure is

(45) S NP kantoku 'director' NP NP haiyuu 'actor' ni V V nak- 'cry' V sase-ta 'cause'

In both (44) and (45), the surface subject receives the nominative marker ga.

When the embedded clause of a causative construction is transitive, the underlying embedded subject receives the ni-marking, as shown below:

(46) Taroo wa Hanako ni hon o yom-ase-ta.
to book   read-cause-Past
  'Taroo caused Hanako to read a book.'
This is a widespread cross-language phenomenon. For example, observe the following French sentences:
(47) a. Underlying J'ai fait [Jean partir]
  b. Surface Sentence J'ai fait partir Jean.
'I made John leave.'
(48) a. Underlying: J'ai laissé [Jean chanter l'hymne]
  b. Surface Sentence J'ai laissé chanter l'hymne a Jean.
'I let John sing the hymn,'
The subject of the complement clause of faire 'make'/laisser 'let' is realized as accusative object in the surface structure if the complement clause is intransitive, and as dative object if it is transitive. See Aissen 1974 for details.

It seems that sentence (46) is ambiguous between the "message transmission" interpretation (i.e., 'Taroo told Hanako to read a book') and the "hands off" interpretation. That the latter interpretation is possible can be seen by the following example:

(49) Taroo wa kodomo ni kaze o hik-ase-te simatta.
child to cold   catch-cause-Cont. ended-up
  'Taroo caused his child to catch the cold (he could or should have prevented it from happening).'
Thus, I hypothesize that (46) is derived from the following two underlying structures:
(50) a. O-causative ("Hands Off")
[Taroo Hanako [Hanako hon yom] sase-ta].
  book read  
  b. Ni-Causative ("Message Transmission")
[Taroo [Hanako hon yom] sase-ta].
For (50a), Equi NP Deletion applies and deletes the underlying embedded clause subject. Next, Verb Raising applies, triggering application of the tree-pruning rules, yielding:
(51)   S NP Taroo NP Hanako NP hon 'book' V V yom-'read' V sase-ta 'cause'
The NP corresponding to Hanako now occupies the indirect object position with respect to the verb complex yom-(s)ase-ta 'caused to read'. Hence, it receives the dative marking ni. For (50a), on the other hand, Agentive-Ni Attachment applies, marking Hanako with ni. Therefore, both for (50a) and (50b), the underlying embedded subject ends up with the ni-marking.

What is peculiar about the transitive causative is the fact that (52b) is totally ungrammatical:

(52) a.
?Kikanzyuu ga tekihei o uti-korosita.
machine-gun   enemy-troop   shoot-killed
'The machine guns killed enemy troops.'
  b.
*Taityoo wa kikanzyuu ni tekihei o
commander   machine-gun   enemy-troop  
uti-koros-ase-ta.
shoot-kill-cause-Past
'The commander caused the machine guns to kill the enemy troops.'
(52a) is artificial in that it has an inanimate subject for a transitive sentence. However, it is at most awkward, and it is passable as a written sentence. On the other hand, the causative version of the sentence is totally ungrammatical. It is clear that the sentence cannot be derived from the underlying ni-causative structure of the pattern of (49b) because there is no message transmission between the commander and the machine guns. However, there does not seem to be anything semantically wrong with the o-causative underlying structure of the pattern of (50a). It seems that the sentence is ungrammatical for the same reason that (38b) is ungrammatical. Namely, the inanimate noun phrase has received the dative marking ni, which represents a secondary agent.

2.3.8. Order of Verbal Elements

We have already seen that, given a sequence of verbal forms, the rightmost element has a higher scope than the rest of the sequence. Thus, the sequence V + Causative + Passive is the passive of V + Causative (i.e. , 'be caused to V'), while the sequence V + Passive + Causative is the causative form of V + Passive (i.e., 'cause to be V-ed'). Similarly, the sequence V + Incipient + Causative means 'cause to begin to V', while the sequence V + Causative + Incipient means 'begin to cause to V'. Whether a given sequence of verbal forms is acceptable or not is determined by various syntactic and semantic factors. For example, observe the following sentences:

(53) a.
Boku wa Hanako o nagur-ase-rare-ta-i.
I   hit-Causative-Passive-Desiderative-Nonpast
'I want to be made to hit Hanako.'
  b.
Boku wa Hanako o nagur-are-sase-te
I   hit-Passive-Causative-Continuative
oita.
kept
'I let Hanako be hit.'
(53a) has the causative morpheme preceding the passive (i.e., nagur-(s)ase-rare- 'be caused to hit'), while (53b) has the same morphemes in reverse order (i.e., nagur-(r)are-sase- 'cause to be hit').

I give below some grammatical and ungrammatical sequences:

(54) a.
tabe-sase-rare-ru 'can cause someone to eat'
eat-Causative-Potential-Nonpast
  b.
nagur-are-ta-i 'be eager to be hit'
hit-Passive-Desiderative-Nonpast
  c.
nagur-rare-ta-gar-are-ru 'be shown a sign of wanting to be hit'
hit-Passive-Desiderative-'show a sign of'-Passive-Nonpast
  d.
*tabe-rare-sase-ru 'cause to be able to eat'
eat-Potential-Causative-Nonpast
  e.
*tabe-ta-sase-ru 'cause to be eager to eat'
eat-Desiderative-Causative-Nonpast
  f.
*tabe-rare-rare-ru 'can be eaten'
eat-Passive-Potential-Nonpast
(54d) is ungrammatical because the causative verb requires an action verb in subordination to it, which rare 'can' is not. (54e) is ungrammatical both for the same reason and because the desiderative ta- forms an adjectival stem, while sase 'cause' requires a verb stem preceding it. The ungrammaticality of (54f) requires explanation. Semantically, there does not seem to be anything wrong with what it is intended to mean. The ungrammaticality seems to be due to the repetition of the homophonous passive and potential morphemes. (Potential forms are produced by attaching rare to vocalic stem verbs and (r)e to consonantal stem verbs. Passive forms are produced by attaching rare to both vocalic and consonantal stem verbs. Thus, with vocalic stem verbs, rare is ambiguous between potential and passive.) In fact, tabe-rare-ru, which has only one rare, can be used for 'can be eaten' (potential of a massive action), as well as for 'be eaten' (passive) and 'can eat' (potential).

2.4. Adverbs

2.4.1. Types of Adverbs

The gerundive form -ku of adjectives and the suppletive gerundive form ni of the copula preceded by nominal adjectives (and sometimes by nouns) are used as adverbial expressions. For example, uma-i 'is good': uma-ku 'being good, well'; akaru-i 'is bright': akaru-ku 'being bright, brightly'; sizuka da 'is quiet': sizuka ni 'being quiet, quietly'; syooziki da 'is honest'; syooziki ni 'being honest, honestly'.

Many Japanese adverbs are of nominal origin: dan-dan '(step-step) gradually'; zen-zen '(all-all) at all'; sono kekka '(that result) as a consequence from it'; sono baai '(that case) in that case'; etc.

Japanese is rich in onomatopoetic adverbs: the following are all onomatopoetic adverbs describing manners of crying: waa-waa 'loudly'; gyaa-gyaa 'noisily'; ogyaa-ogyaa (of a new-born baby); siku-siku 'sobbingly'; meso-meso 'effeminately.'

Many adverbs are optionally or obligatorily followed by the particle to: doo-doo to 'grandly'; seizen to 'in an orderly way'; yukkuri: yukkuri to 'slowly'; honnori: honnori to 'faintly'.

2.4.2. Sentential Adverbs

The gerundive form -ku of adjectives, followed by an exclamatory particle mo, and the suppletive form ni of the copula, preceded by nominal adjectives or nouns, are also used for sentential adverbs.

(1) a.
Yamada- sensei ga onakunari ni natta koto
  teacher   dying being became fact-that
wa kanasi-i.
  is-sad
'That Professor Yamada has died is sad.'
  b.
Kanasi-ku mo, Yamada-sensei ga onakunari ni natta.
'Sadly, Professor Yamada has died.'
(2) a.
Yamada- sensei ga onakunari ni natta koto
  teacher   dying being became fact-that
wa kanasii koto da.
  is-sad thing is
'That Professor Yamada has died is a sad thing.'
  b.
Kanasii koto ni, Yamada-sensei ga onakunari ni natta.
(Lit.) '(it) being a sad thing, Professor Yamada has died.' 'To our sorrow, ...'
Sentential adverbs with -ku mo are used primarily in formal and literary writing and seldom enter into colloquial speech.

I give below representative samples of sentential adverbs:

(3) Adjectival
 
osorosi-i
'is dreadful'
osorosi-ku mo
'dreadfully'
 
yorokobasi-i
'is joyful'
yorokobasi-ku mo
'happily'
 
nagekawasi-i
'is regrettable'
nagekawasi-ku mo
'regrettably'
(4) Nominal-Adjectival
 
tasika da
'is certain'
tasika ni
'certainly'
 
hontoo da
'is true'
hontoo ni
'truly'
 
hukoo da
'is unfortunate'
hukoo ni mo
'unfortunately'
 
saiwai da
'is fortunate'
saiwai ni (mo)
'fortunately'

2.4.3. Some Characteristics of Adverbs

In Japanese, adverbs are placed to the left of the constituents that they modify:

(5) a.
Totemo omosiroi hon o yonda.
very is-interesting book   read
'I read a very interesting book.'
  b.
Motto yukkuri aruke.
more slowly walk-Imper.
'Walk more slowly.'
In colloquial speech, adverbs can appear postverbally as "afterthoughts" (see section 2.1.2).
(6) a.
Kimi wa hontoo ni baka da ne.
you   truly   stupid   isn't-it
'You are truly stupid.'
  b.
Kimi wa baka da ne, hontoo ni.
(7) a.
Taroo ga mata kita yo.
  again came I-am-telling-you
'Taroo has come again.'
  b.
Taroo ga kita yo, mata.
(8) a.
Motto yukkuri aruke. = (5b)
more slowly walk-Imper.
'Walk more slowly.'
  b.
*Aruke, motto yukkuri.
(9) a.
Hayaku koko ni koi.
fast here to come-Imper.
'Come here quickly.'
  b.
Koko ni koi, hayaku.
In the above, (8b) is unacceptable because the first part of the sentence (i.e., Aruke 'Walk!') orders the hearer to walk as opposed to running or standing still, while the postverbal element (i.e., motto yukkuri 'more slowly') presupposes that the hearer has already been walking.

In many languages, adpositional phrases are ambiguous between the adjectival and adverbial use. For example, the following sentence is ambiguous:

(10) John called up the visitor from Iowa.
The sentence can mean either 'From Iowa, John called up the visitor' or 'John called up the visitor who was from Iowa.' Ambiguity of this type does not exist in Japanese, in which adjectival postpositional phrases are always marked with no, the prenominal attributive form of the copula.
(11) a.
Taroo wa Aiowa kara kyaku ni denwa o kaketa.
  Iowa from visitor to phone   placed
'From Iowa, Taroo called up the visitor.'
  b.
Taroo wa Aiowa kara no kyaku ni denwa o kaketa.
  Iowa from is visitor to phone   placed
'Taroo called up the visitor who was from Iowa.'

2.5. Compound and Complex Sentences

2.5.1. Coordination and Subordination

Two clauses are combined together in coordination or subordination with the use of the gerundive, continuative, or conditional form of verbals, particles, or formal nouns. Observe the following sentences:

(1) Gerundive
  Taroo ga Amerika ni ik-i, Hanako ga Huransu ni itta.
    to go-Ger.   France to went
  'Taroo went to America and Hanako went to France.'
(2) Continuative
  Taroo ga Amerika ni it-te, Hanako ga Huransu ni itta.
    to go-Cont.   France to went
  'Taroo went to America, and Hanako went to France.'
(3) Particle
  a.
Taroo ga Amerika ni itta si, Hanako ga Huransu
  to went and   France
ni itta.
to went
'Taroo went to America, and Hanako went to France.'
  b.
Taroo wa hon o katta si, rekoodo mo katta.
  book   bought and record   bought
'Taroo bought books, and (he) also bought records.'
(The noun-coordinating particle to cannot be used for coordinating clauses.) The above are all examples of coordinated clauses. The three types are different in a very subtle way. (1) is neutral coordination, while (2) represents a temporal or logical sequence (i.e., 'and then, and therefore'). (3) implies that other things also took place. Thus (3b) means that Taroo did many things such as buying books and buying records.

The following sentences illustrate various types of subordination:

(4) Gerundive
  a.
Bukka ga agar-i, minna ga komatte iru.
price   rise-Ger. all   suffering are
'Prices rising, all are suffering.'
  b.
Taroo wa tosyokan ni ik-i, hon o yonda.
  library to go-Ger. book   read
'Going to the library, Taroo read books.'
(5) Continuative
  a.
Bukka ga agat-te, minna ga komatte iru.
price   rise-Cont. all   suffering are
'Prices having risen, all are suffering.'
  b.
Taroo wa, tosyokan ni it-te, hon o yonda.
  library to go-Cont. book   read
'Having gone to the library, Taroo read books.'
(6) Conditional
 
Kono siken wa gozyutten tor-eba, pasu dekiru.
this test   50-points take-if pass can
'You can pass this test if you get 50 points.'
(7) Particles
  a.
Bukka ga agatta node, minna ga komatte iru.
price   rose since all   suffering are
'Because prices have gone up, all are suffering.'
  b.
Taroo wa, tosyokan ni itta noni, hon o yomanakatta.
  library to went though book   didn't-read
'Taroo, although he went to the library, didn't read books.'
  c.
Taroo ga gakkoo ni iku to, Hanako ga matte ita.
  school to go when   waiting was
'When Taroo got to school, Hanako had been waiting.'
(8) Formal Nouns
  a.
Taroo ga kita toki, Hanako wa mada nete ita.
  came time   still sleeping was
'(At the) time when Taroo came, Hanako was still asleep.'
  b.
Kono-aida Tookyoo ni itta sai, Yamada- sensei ni
other-day   to went case   teacher to
omenikakatta.
met-Honorific
'I met Teacher Yamada when I went to Tokyo the other day.'

Of particular interest in the above examples is the fact that the gerundive and continuative forms of verbals can be used for both coordination and subordination. Which of the two types is involved in a given sentence depends upon the semantics of the two clauses and their relationship with each other, (1, 2) and (4, 5) show different behaviors with respect to various syntactic processes. For example, the subject of the second clause in (4, 5) can be relativized, but that of (1, 2) cannot.

(9) a.
*[Taroo ga Amerika ni ik-i, Huransu itta] hito
  to go-ing France went person
*'a person such that Taroo went to America and (he) went to France'
  b.
[Bukka ga agar-i, komatte iru] hito-tati
price   rise-ing suffering are persons
'people who, prices going up, are suffering'
Similarly, the interrogative particle can be attached to (4, 5) but not to
(10) a.
??Taroo ga Amerika ni ik-i, Hanako ga Huransu
  to go-ing   France
ni itta ka.
to went Q
'Did Taroo go to America and Hanako to France?'
  b.
Bukka ga agar-i, minna komatte iru ka.
price   rise-ing all suffering are Q
'Prices going up, are all suffering?'
Instead, one has to nominalize the entire sentence first, and embed it in the x da '(it) is x' pattern:
(11)
[Taroo ga Amerika ni ik-i, Hanako ga Huransu ni itta]
to go-ing   France to went
 
no ka.
that Q
In the above example, da has been deleted before ka by the rule given in section 2.1.3.3.

Likewise, the subject of the second clause of (4, 5) can be a wh-interrogative word, but not that of (1, 2):

(12) a.
??Taroo ga Amerika ni ik-i, dare ga Huransu
  to go-ing who   France
ni itta ka.
to went Q
(Lit.) 'Taroo went to America, and who went to France?'
  b.
Bukka ga agar-i, dare ga komatte iru ka.
price   rise-ing who   suffering is Q
'Prices going up, who is suffering?'

2.5.2. Indirect Statements and Questions, Quotatives

Subject and object complements are marked with the formal noun koto (Lit., 'thing') 'that, the fact that', the nominalizer no 'that', or the postpositional particle to 'that'. The first two are followed by case-marking particles, but to is not. Roughly speaking, koto and no are used for factive complements and to for nonfactive complements.

(13) a.
Taroo ga kekkonsita koto wa hontoo da.
  married that   true is
'It is true that Taroo got married.'
  b.
Taroo ga kekkon tyokugo sinde simatta
  marriage right-after dying ended-up
no wa higeki da.
that   tragedy is
'It is tragic that Taroo died right after he got married.'
(14) a.
Boku wa Taroo ga kekkonsita koto o kiita.
I   married fact-that   heard
'I heard about the fact that Taroo got married.'
  b.
Boku wa Taroo ga kekkonsita to kiita.
I   married that heard
'I heard that Taroo got married.'
  c.
Boku wa Taroo ga piano o hiku no o kiita.
I   play that   heard
'I heard Taroo play the piano.'
The difference between koto 'the fact that' and no 'that' lies in the degree of abstraction. Koto represents an abstract fact, while no represents an unabstracted concrete action or state. See Kuno 1973, Ch. 18, and Josephs 1972 for details.

Indirect questions are formed by embedding questions (ending with the question particle ka) without complementizers:

(15) a.
Boku wa Taroo ga kekkonsita koto o sitte iru.
I   married fact-that   knowing am
'I know that Taroo got married.'
  b.
Boku wa Taroo ga dare to kekkonsita ka
I   who with married Q
sitte iru.
knowing am
'I know whom Taroo married.'
Indirect question formation and the use of complementizers interact with each other in a complex fashion. Compare the following sentences:
(16) a.
Taroo wa Hanako ga kuru to itta.
  come that said
'Taroo said that Hanako was coming.'
  b.
Taroo wa Hanako ga kuru to itta ka.
'Did Taroo say that Hanako was coming?'
  c.
Taroo wa Hanako wa kuru ka to itta.
'Taroo said, "is Hanako coming?"'
  d.
Taroo wa Hanako wa kuru ka to itta ka.
'Did Taroo say, "is Hanako coming?"'
  e.
Taroo wa dare ga kuru ka itta.
  who   come Q said
'Taroo said who was coming.'
  f.
Taroo wa dare ga kuru ka itta ka.
'Did Taroo say who was coming?'
  g.
Taroo wa dare ga kuru to itta ka.
'Who did Taroo say was coming?'
  h.
Taroo wa dare ga kuru ka to itta.
'Taroo said, "Who is coming?"'
  i.
Taroo wa dare ga kuru ka to itta ka.
'Did Taroo say, "Who is coming?"'
An interrogative word (e.g., dare 'who') is bound by the ka which is closest to it on its right side. A question in which ka has an interrogative word that it binds is an interrogative-word question, and one in which ka does not have an interrogative word is a yes-or-no question. If a complement clause is neither an interrogative-word question nor a yes-or-no question, it retains its own complementizer. The above three principles explain the semantics of all the sentences in (16).

2.6. Grammatical Processes

2.6.1. Pronouns and Deletion of Noun Phrases

Japanese lacks authentic pronouns for any grammatical persons. Most existing forms that correspond to pronouns in other languages are derived from nominal expressions: boku '(your) servant — I', watakusi 'personal — I', kimi 'lord — you', anata 'far away — you', omae 'honorable (person in) front (of me) — you', kare 'thing far away — he', kanozyo 'far away woman — she', karera 'far away + Plural — they'. Third person pronouns are used only in pedantic speech by educated people. Different forms of the first and second person pronouns are used depending upon the relative status of the speaker and the hearer, and upon the speech levels. The following give representative samples:

(1) First Person Second Person
  watakusi polite
  watasi quasi-polite anata quasi-polite (see below); informal when used by females
  boku informal
(limited to male)
kimi informal (generally used only by males addressing males)
  ore vulgar
(limited to male)
anta slightly vulgar
omae vulgar
temae extremely vulgar

Japanese lacks polite second person pronouns that can be used when addressing superiors. Otaku 'your honorable house — you' comes closest to such a pronoun, but it is seldom used when addressing one's superiors. The speaker uses either the addressee's proper name plus title or title only or, alternatively, he resorts to deletion when the second-person reference is clear. I will illustrate this point with examples:

(2) Child to Mother
  {0 ?Watakusi ('I')} {0 okaasan ('mother') ni} }
 *anata ('you') to' }
 *otaku ('you') }
 
kore agemasyoo ka.
'this give-will Q'
 
'Shall I give this to you?'
(3) Father to Small Daughter
  { Papa Otoosan ('father') ??Boku ('I') *Watakusi ('I') }
 to kaimono ni ikoo. 'with shopping for let's-go'
 
'How about going shopping with me?'
(4) Student to Teacher Yamada
  { *Anata ('you') *Otaku ('you') Sensei ('teacher')
 Yamada-sensei } ni to o-kik-i sitai 'Honorific-ask-ing do-want
 
koto ga arimasu.
thing   there-is'
 
'I have something that I would like to ask you about.'
(5) To Mr. Yamada
  ?Anata ('you') Otaku ('you') Yamada-san ('Mr. Y.') wa doko
 no 'where 's
 
go-syussin desu ka.
Honorific-origin is Q'
 
'Where do you come from?'
 
b. Kimi ('you') wa doko no syussin kai. (Informal)
  0   where 's origin Q
  'Where do you come from?'

Deletion is based on discourse recoverability. Observe the following sentences:

(6) a.
Taroo ga Koobe ni kita node, 0 0 ai ni itta.
  to came since   see to went
'Since Taroo came to Kobe, (I) went to see (him).'
  b.
*0 Koobe ni kita node, 0 Taroo ni ai ni itta.
  to came since   to see to went
'Since (he) came to Kobe, (I) went to see Taroo.'
In (6a), the main clause subject 'I' is recoverable because it refers to the speaker, which is the most discourse-presupposed of all nouns in general. The object of ai 'to see' is also recoverable because Taroo has already been mentioned in the node clause. Hence, the acceptability of the sentence. On the other hand, (6b) is unacceptable because the subject of the node clause, barring a mention of his name in the immediately preceding discourse, is not recoverable from left context.

Observe, further, the following sentence:

(7)
Yamada- kyoozyu no ronbun o gakusei ga eigo
  Prof. 's paper   student   English
 
ni honyakusita.
to translated
 
'A student translated Prof. Yamada's paper into English.'
Gakusei 'student', in this sentence, is in fact ambiguous between 'a student', 'my student', and 'Professor Yamada's student'. To state clearly that Professor Yamada's student translated the paper, one says the following:
(8)
Yamada- kyoozyu no ronbun o kyoozyu no gakusei
  Prof. 's paper   Prof. 's student
 
ga honyakusita.
  translated
 
(Lit.) 'Professor Yamada's paper, the professor's student translated.'
In (8), kyoozyu 'professor' is used in place of a third person pronoun.

2.6.2. Deletion of Verbs

English has a syntactic process called Verb Phrase Deletion that deletes a verb phrase, leaving behind an auxiliary verb or infinitival to. For example, observe the following sentences:

(9) Speaker A: Did you go to Boston yesterday?
  Speaker B: Yes, I did. (go to Boston yesterday deleted)
(10) Speaker A: Who was killed in the accident?
  Speaker B: John was. (killed in the accident deleted)
(11) Speaker A: Do you want to go there?
  Speaker B: Yes, I want to. (go there deleted)
Japanese does not have an auxiliary verb that can be used independently. Therefore, a rule of Verb Phrase Deletion is nonexistent in Japanese. This situation should result in excessive redundancy, but Japanese copes with this situation in two ways. First, it uses the template x da to give the minimal answer:
(12) Speaker A:
Dare ga kinoo gakkoo ni kimasen
who   yesterday school to come-not-Polite
desita ka.
was-Polite Q
'Who didn't come to school yesterday?'
  Speaker B:
Taroo desu.
(Lit.) 'Is (Polite) Taroo.'
The second device that the language uses is to repeat the main verb, deleting everything else that is recoverable from context. For example, observe the following exchanges:
(13) Speaker A:
Kimi wa Hanako ni moo tegami o
you   to already letter  
kakimasita ka.
wrote Q
'Have you already written a letter to Hanako?'
  Speaker B:
Hai, kakimasita.
yes wrote
'Yes, (Lit.) (I) wrote.'
(14) Speaker A:
Kimi wa Amerika ni itta koto ga
you   to went experience  
arimasu ka.
have Q
'Have you been to America?'
  Speaker B:
Hai, arimasu.
'Yes, (I) have.'
(15) Speaker A:
Kimi wa Amerika ni ikitai to omotte
you   to go-want that thinking
imasu ka.
are Q
(Lit.) 'Are you thinking that you want to go to America?'
  Speaker B:
Hai, omotte imasu.
'Yes, (Lit.) (I) am thinking.'
Copulas cannot have their precopular predicate missing. One either has to repeat the precopular predicate or say 'It is so,' as shown below:
(16) Speaker A:
Tanaka-san wa syoogakkoo no sensei
  grade-school 's teacher
desu ka.
is Q
'Is Mr. Tanaka a grade-school teacher?'
  Speaker B:
*Hai, desu.
'Yes, (he) is.'
Hai, syoogakkoo no sensei desu.
  grade-school 's teacher is
'Yes, (he) is a grade-school teacher.'
Hai, soo desu.
  so is
'Yes, (it) is so.'
Japanese, like many other languages, has a process called Right-Node Raising, which extracts the rightmost common constituent from conjuncts. This process applies to the structure underlying (17a), and yields (17b):
(17) a.
Taroo ga Amerika ni ik-i, Hanako ga Huransu
  to go-ing   France
ni itta.
to went
'Taroo went to America, and Hanako went to France.'
  b.
Taroo ga Amerika ni, Hanako ga Huransu ni, itta
  to   France to went
'Taroo went to America, and Hanako, to France.'

On the other hand, Japanese lacks a rule of gapping, which deletes from conjuncts (except for the leftmost one) verbs that are repeated:

(18) a. John went to Japan, and Mary went to France.
  b. John went to Japan, and Mary 0 to France.
The following sentence, which would derive if Gapping had applied to the structure of (17a), is ungrammatical:
(19)   *Taroo ga Amerika ni {ik-i, it-ta} Hanako ga Huransu ni 0.

2.6.3. Scrambling

As I have illustrated with examples at the beginning of this paper, word order in Japanese sentences is relatively free. Sometimes, it is possible to move a constituent in a subordinate clause to sentence-initial position. Observe the following examples:

(20) a.
Boku wa [Yamada- sensei ni anata o syookaisitai]
I   teacher to you   introduce-want
to omotte imasu.
that thinking am
(Lit.) 'I have been thinking that I want to introduce you to Teacher Yamada.' 'I want to ...'
  b.
Anata o boku wa [Yamada-sensei ni syookaisitai] to omotte imasu.
(21) a.
Boku wa [[Yamada- sensei ni anata o syookaisitai]
I   teacher to you   introduce-want
to yuu kiboo] o motte imasu.
that say hope  
'I have been entertaining the hope that (Lit.) I want to introduce you to Teacher Yamada.'
  b.
Anata o boku wa [[Yamada-sensei ni syookaisitai] to yuu kiboo] o motte imasu.
In (20b), anata o has been fronted from a reportive to clause; while in (21b) the same expression has been fronted from a complex NP of the type the hope that. The following example shows that scrambling out of a subordinate clause is subject to some kind of syntactic or semantic constraint. (See Haig 1976 for further discussion of this topic.)
(22) a.
Taroo wa [[Hanako to Amerika ni iku] keikaku]
  with   to go  
o toriyameta.
  canceled
'Taroo canceled (his) plan to go to America with Hanako.'
  b.
*Amerika ni Taroo wa [[Hanako to iku] keikaku] o toriyameta.

2.6.4. Foregrounding: Theme

As discussed briefly in section 2.1.6, Japanese has a particle specifically used for marking the theme of a sentence. Time and place adverbs that are used for setting up the scene for the rest of the sentence readily serve as themes. Among nominal constituents such as subjects, objects, and datives, subjects qualify as themes most easily of all. However, other constituents of the sentence can also qualify for themes:

(23) a.
Taroo wa gakusei desu.
  student is
'Taroo is a student.'
  b.
Kyoo wa boku wa tyuusyoku wa nuki ni siyoo.
today   I   lunch   skipping -ly do-will
'Today, I will skip lunch.'
  c.
Tukue no ue ni wa hon ga sansatu atta.
desk 's top on   book   3-volume was
'On the desk, there were three books.'
  d.
Kono hon wa dare ga kaita ka wakaranai.
this book   who   wrote Q understand-not
'This book — we don't know who wrote (it).'
In (23a), the wa-marked theme is coreferential with the underlying subject of the sentence. In (23b), kyoo wa 'today' is a scene-setting thematic time adverb; boku wa is coreferential with the underlying subject of the nominalized verb nuki 'skip', and tyuusyoku wa, with its underlying object. In (23c), the theme is a scene-setting thematic place adverb. What is most interesting is the grammaticality of (23d). Kono hon wa 'this book' is coreferential with the object of the verb of the embedded interrogative clause. Similarly, observe the following sentence:
(24)
Tanaka- sensei wa, hon o syuppansuru koto ni
  teacher   book   publish fact-that being
 
natte ita syuppansya ga toosansite simatta.
becoming was published   bankrupt went
 
(Lit.) 'Professor Tanaka, the publisher that was scheduled to publish (his) book went bankrupt.'
In (24), the theme of the sentence is coreferential with the underlying genitive (modifying hon 'book') in a relative clause. On the other hand, the following sentence is unacceptable:
(25)
??Tanaka- sensei wa hon o syuppansuru koto ni
  teacher   book   publish to -ly
 
natte ita syuppansya ga syain o hyakunin
becoming was publisher   employer   100-people
 
saiyoosita.
employed
 
(Lit.) 'Professor Tanaka, the publisher that was scheduled to publish (his) book has employed one hundred employees.'
As the above examples show, whether a given thematic sentence is acceptable or not depends not so much on syntax, as on the semantic relationship that holds between the theme and the rest of the sentence. In (24), for example, one can establish a semantic relationship between Professor Tanaka and the publisher's having gone bankrupt. Perhaps the bankruptcy of the publisher affected Professor Tanaka in that his book could not be published. (24) is acceptable to native speakers of Japanese to the extent that such a semantic relationship can be established. On the other hand, it is difficult to establish a connection between Professor Tanaka and the fact that the publisher has hired one hundred new employees. Hence results the unacceptability of (25). It seems that all we can say at present is that in a thematic sentence, the rest of the sentence must be a statement about the theme. The conclusion that it is not possible to define any syntactic relationship that must hold between the theme and the rest of the sentence is also confirmed by the fact that, as briefly discussed in section 2.1.6, there are thematic sentences whose themes do not have any underlying syntactic functions, as in the following sentences:
(26) a.
Sakana wa tai ga ii.
fish   redsnapper   is-good
(Lit.) 'Speaking of fish, the redsnapper is the best.'
  b.
Basyo wa okunai-setu ga attooteki datta.
place   indoor-theory   predominant was
'Speaking of the place (where the crime took place), the indoor-theory was predominant.'

Theme in Japanese does not interact with very many transformational processes. It is the subject, and not the theme, that triggers Reflexivization. (See section 2.3.3. Kuno 1976c shows that, in rare circumstances, it is possible to use the theme of a sentence as trigger for Reflexivization.) It is the subject, and not the theme, that triggers the honorific marking on verbs of the type discussed in section 2.1.4. For example, toosansite simatta 'went bankrupt' of (24) cannot receive honorific marking with respect to Tanaka-sensei because the latter is the theme, and not the subject, of the sentence. It is the subject, and not the theme, that undergoes Subject Raising. (See Kuno 1976a for Subject Raising in Japanese. The fact that the subject, but not the theme, plays a predominant role in processes such as Reflexivization, Passivization, Equi NP Deletion, etc., has been observed in Li and Thompson 1976b.)

2.6.5. Clefting

Examples of cleft sentences are given below:

(27) a.
Hanako to kinoo Koobe ni itta no wa Taroo da.
  with yesterday   to went that   is
'It was Taroo who went to Kobe with Hanako yesterday.'
  b.
Taroo ga kinoo Koobe ni itta no wa Hanako to da.
'It was with Hanako that Taroo went to Kobe yesterday.'
  c.
Taroo ga Hanako to kinoo itta no wa Koobe da.
'It was Kobe that Taroo went (to) with Hanako yesterday.'
  d.
Taroo ga Hanako to Koobe ni itta no wa kinoo da.
'It was yesterday that Taroo went to Kobe with Hanako.'
Japanese cleft sentences lack a dummy subject corresponding to the English it, and require the main clause verb (copula) to be in the nonpast tense. Lack of It-Clefting (as well as of It-Extraposition) is a characteristic of SOV languages (see Kuno 1974) and of topic-prominent languages (see Li and Thompson 1976b).

The nonpast tense da of the main clause shows that the judgment represented in the cleft sentence is made at the time of the utterance. If the past tense copula datta 'was' is used, it shows that the judgment was made in the past.

Note

Research represented in this paper has been supported in part by the National Science Foundation's grant to Harvard University (Grant No. SOC-7412366). I am greatly indebted to Winfred Lehmann for many valuable suggestions on the organization of the paper. I am also grateful to Linda Shumaker, Ruth Stevens, and Kazue Campbell for many helpful comments on the earlier version of the paper.