Chapter 3 THE FRAMEWORK AND LANGUAGES THAT SUPPORT IT BY THEIR MERGERS: JAPANESE1

Chapter 3 THE FRAMEWORK AND LANGUAGES THAT SUPPORT IT BY THEIR MERGERS: JAPANESE1 1. Introduction Japanese is a topic prominent SOV language. Predicate adjectives occur in the final V position without a copula (as in [1]), and when they are intended in a modifying CLASSIFICATION sense, they are positioned before their modified partner (as in [2]) (Kuno 1978.66): (1) Taroo ga mada waka-i [Taroo still young-pres] Taroo is still young (2) Waka-i hito ga ki-ta [young-pres person come-past] The young person came The postpositions ga (cf. [1] & [2]) and wa (cf. [3] below]) are the muchdiscussed Japanese subject and topic markers, respectively. The postposition o generally marks patients; and ni marks recipients and locations (among other uses). Cf. Davis 1990 for a general interpretation of their senses (in which ga is not subject, and in which ni is a marker of PERIPHERAL content). 2. The use of no: relative clauses and adjectives Japanese restrictive relative clauses look much like adjectival clauses. The modifying restrictive relative clause is preposed to its syntactic head and retains its grammatical accoutrements; it lacks the equivalent of the English relative pronoun and the verb form that appears at the end of a relative clause is identical with that of a sentence-final verb (Kuno 1973.146, 234 & 235): 1 I would like to thank Jack Wiedrick for comments on this chapter.

52 THE SEMANTICS OF SYNTACTIC COMPLEXITY (3) (a) John wa hon o yon-da [John book read-past] John read the book (b) John wa,yon-dahon o katazuke-ta [ read-past book put.away-past] John put away the book that he read (4) (a) Watakusi ga Mary o sit-te iru [I know-gerund Copula] I know Mary (b) Watakusi ga sit-te iru Mary [I know-gerund Copula ] the Mary that I know Kuno (1973.30-31) describes Japanese verbs as actions and not states ; and in the present tense, verbs are felt as indicating a semantically future time (Kuno 1973.136-37). To express a present time, e.g. knows in (4), action verbs must be followed by i-ru be in the state of (Kuno 1973.30). Thus, we have (5) (Kuno 1973.137 & 146): (5) (a) John ga kono hon o yom-u [ that book read-pres] John will read this book (b) John wa,yom-u hon o katazuke-ta [ read-pres book put.aside-past] John put aside the book that he was about to read Relative clauses take two distinctive shapes, but with no contrast between restrictive and non-restrictive. Compare (3) - (5) with (6), which employs a no (Kuno 1973.73):2 2 Jack Wiedrick makes the following remarks about (6): This example is not necessarily a non-restrictive relative clause. Cp. this exchange: A: dare ga itiban yoku dekiru daroo? who GA best well be.able probably I wonder, who does the best?

JAPANESE 53 (6) Kono class no John ga yoku dekiru [this class well is-able] John, who is in this class, does well The dependent clauses in (3b), (4b), (5b), and (7) represent Type 1 nonrestrictive relative clauses (Kuno 1973.235): (7) Watakusi ni eigo o osie-te iru Mary [I English teach-gerund Copula ] Mary, who is teaching me English The relative clause in (6) is a Type 2 non-restrictive relative clause. The element no, which appears in (6), has a restrictive use, but it seems to be B: sore wa Johndaroo that WA J. probably That would probably be John A: demo futari iru daroo? asa no kurasuno to ima no but 2.person exist probably morning NO class NO and now NO kurasu no class NO But aren t there two? One in the morning class, and one in (our) class now B: soo datta ka? thuswas? Oh, were there? A: soo da yo. dakara, doti? thusis EMPH therefore which Yes, there are. So, which one? B: kono kurasu no John ga yoku dekiru this class NO J. GA well be.able The one in this class (is the one who) does best It seems as if Kuno simply wanted to make a point, and chose to see only the interpretation that suited him. I [JW] personally feel that the English-type restrictive/nonrestrictive contrast is one that doesn t really apply to Japanese syntax. Consider: gakusei no John student NO J. John, the student John the student the student, John the student John How it should be glossed depends on the context in which you find it.

54 THE SEMANTICS OF SYNTACTIC COMPLEXITY limited to constructs involving numbers and classifiers (Kuno 1973.25): (8) (a) ni-satu no hon [two-volume book] two books (b) ip-pai no mizu [one-cup water] a cup of water Kuno (1973.25. Cf. also Kuno 1973.74) says of such: The no in these examples [i.e. (8)] is not the genitive particle no but the attributive form of the copula da, which appears in... (i) John wa gakusei da student is John is a student (ii) gasukei no John student is John, who is a student Compare (6) with (9), for example, in which the no is, apparently, the mark of possession (Kuno 1973.68): (9) Kono class no dansei ga yoku dekiru [this class s male well are-able] This class s boys (and only they) do well The use with numbers and classifiers as in two-volume book appears to be the only restrictive use of no. Other uses pattern as does the non-restrictive (6). Candidates, which like (9), may expand no into a restrictive use are declared to be like (9) examples of the genitive no. Except for the distinction between John and dansei, the two sentences are identical, and we might suppose that there is one syntactic form here, and not two. Recall Farsi - e; the good horse and asb-e mard the man s horse. Mandarin also blends the possessive mark with the modificational: the

JAPANESE 55 teacher s car and the thief who is running.3 2.1 Complementisers: the use of no and to The shape no occurs still again as a complementiser (This is Kuno s term.) along with to (Kuno 1973.213): (10) (a) Watakusi wa John ga Mary o butu no o mi-ta [I hits see-past] I saw John hitting Mary (b) *Watakusi wa John ga Mary o butu to mi-ta [I hits see-past] (11) (a) *John wa nihongo ga muzukasii no o it-ta [ Japanese difficult-is say-past] (b) John wa nihongo ga muzukasii to it-ta [ Japanese difficult-is say-past] John said that Japanese is difficult According to Kuno (1973.213 & 215):... no clauses represent an action, state, or event that the speaker presupposes to be true, while the to clause represents an action, state, or event that does not have such a presupposition... it is natural that to, which was originally a particle for reporting someone else s statement, be used for representing an action, state, or event about which the speaker has not made a presupposition. The suggestion from this observation is that the relation of the contents of the PROPOSITIONS connected by the complementiser no is a tighter one than the relation between the contents of PROPOSITIONS connected by to. The postposition no is used in the expression of possessives, adjectives, relative clauses, and as a complementiser. Kuroda (1976.270), in discussing certain relative clause constructions using no, 3 Tsubomoto (1976.400) notes that in A no B combinations, where the sense is possessive, that In genitive A no B construction [sic], A is generally larger than B in the semantic domain, as in nihon no Tokyo Tokyo of Japan rather than *Tokyo no nihon Japan of Tokyo.

56 THE SEMANTICS OF SYNTACTIC COMPLEXITY (12) Taroo wa ringo sara no ue ni at-ta no o [Taroo apple plate on be-past tot-te, poketto ni ire-ta take-gerund pocket put.in-past] Taroo picked up an apple which was on a plate and put it in a pocket (13)?Taroo wa Hanako ga kinoo ringo o kat-ta noo [Taroo Hanako yesterday apple buy-ger tot-te, poketto ni ire-ta take-gerund pocket put.in-past] Taroo picked up an apple which Hanako bought yesterday and put it in a pocket remarks that it would be a piece of fortuitous good luck if one s informant accepted (13).4 The difficulty seems to lie in the lack of connection between Hanako s buying an apple and Taroo s picking it up and putting it in his pocket. The first act is irrelevant to the second and that relevancy is precisely what is required with/signalled by no (Kuroda 1976.270): THE RELEVANCY CONDITION: For a headless relative clause to be acceptable, it is necessary that it be interpreted pragmatically in such a way as to be directly relevant to the pragmatic content of its matrix clause... Note that in order for the above justification for direct relevancy of... [Ringo ga sara no ue ni atta There was/were an apple/apples on a plate/plates ] in... [(12)] to be valid, it is necessary that... [Ringo ga sara no ue ni atta] be interpreted as simultaneous with the time reference of the matrix clause. Tsubomoto (1981) extends the relevancy condition to the contrasts of (10) and (11) and also to the cleft -looking sentences of (14) (Tsubomoto 1981.396): (14) (a) Ringo o tabe-ta no wa Taroo da [apple eat-past Taroo be] It s Taroo that ate an apple 4 Jack Wiedrick remarks that This sentence would work better rearranged and with a pause added : (i) Taroo wa ringo, Hanako ga kinoo katta no o totte, poketto ni ireta. [T. WA apple H. GA yesterday bought NO O taking pocket in inserted Taroo took the apple, that Hanako bought yesterday, and put it in his pocket

JAPANESE 57 (b) *Ringo o tabe-ta to wa Taroo da [apple eat-past Taroo be] It s Taroo that ate an apple In (10), no succeeds because the speaker s act of witnessing and the event which is witnessed (John hitting Mary) can be amalgated as component aspects of a larger occurrence.5 In (11), the speaker s act of claiming and the fact that he claims (that Japanese is difficult) are unconnected; they share no common world, hence, no fails. The element to then marks the looser relation of (11), and it in turn fails in the tighter relationship between the events in (10). The common thread which unites the usages of no is expressed by Tsubomoto (1981.395): UNIFYING FUNCTION OF no: No unifies two propositions closely, as if they were one event This broader interpretation is implied by Kuroda (1976.273) with his examples in (15): (15) (a) Taroo wa Hanako ga osoikakat-te [Taroo Hanako approach.to.attack-gerund ki-ta no o nejihuse-ta come-past floor.and.hold.down-past] Taroo floored and held down Hanako, who had approached to attack (b)?taroo wa Hanako ga harubaru tazunet-te [Taroo Hanako a.long.way visit-gerund ki-ta no o nejihuse-ta come-past floor.and.hold.down-past] Taroo floored and held down Hanako, who had come a long way to visit him The problem with (15b) is that one cannot plausibly see Hanako s coming a long way to visit and Taroo s flooring her and holding her down as reasonable aspects of a coherent whole ( visit... floor.hold.down ); in (15a), the coherency is apparent ( approach.to.attack... floor.hold.down ). Therefore, 5 Both Tsubomoto (1981.393) and Kuroda would see simultaneity as the necessary condition supporting (10).

58 THE SEMANTICS OF SYNTACTIC COMPLEXITY (15a) succeeds where (15b) does not. 2.2 A third complementiser: koto A third complementiser koto is used in Japanese. It, like no, indicates that the speaker presupposes its clause to be true (Kuno 1973.213 & 215-16): (16) John wa Mary ga baka da to nagei-ta [ stupid is deplore-past] John deplored that Mary was stupid In (16), phrased with to, Mary may or may not, in fact, be stupid; but in (17), the presence of koto implies that she indeed is: (17) John wa Mary baka na koto o nagei-ta [ stupid is deplore-past] John deplored that Mary was stupid Because the verbs hayaigattensita formed a hasty conclusion, itta 'said, omotta thought, kangtigaisita made the wrong guess, and gokaisita formed the wrong notion identify events whose complements are necessarily uncertain, they may appear only with to and not with koto nor no (Kuno 1973.215 & 217). Further complementary and partially complementary distrubutions of these complementisers exist (Kuno 1973.219-20): Verbs of perception [e.g. see, hear, feel, etc.] can take only no clauses... Verbs of ordering [e.g. order, demand, ask, force, etc. (Kuno 1973.220)] can take koto clauses, but not no clauses... With verbs of expecting it is pereferable to use koto, but no and to are also acceptable. The verb manabu learn also requires koto while rejecting no and to (Kuno 1973.214): (18) (a) Watakusi wa nihongo ga muzukasii koto o manan-da [I Japanese difficult learn-past] I learned that Japanese is difficult (b) *Watakusi wa nihongo ga muzukasii no o manan-da (c) *Watakusi wa nihongo ga muzukasii to o manan-da

JAPANESE 59 We can understand why (18b) is not acceptable; John s learning, as an action, has nothing to do with the fact (or not) that Japanese is difficult. With respect to (18c), the verb manabu learn conflicts with to precisely in the way verbs such as hayagattensita formed a hasty conclusion do not.6 The greater tenuousness between the two events (one of which is form a hasty conclusion ) is compatible with to, whereas in (18c), the closer connection of necessity implied between the two events (one of which is manabu learn ) conflicts with the looseness of to. Sentences (19) - (21) illustrate the possibility of a three-way contrast (Kuno 1973.220): (19) Mary wa John ga kuru no o kitaisite-ita [ come was-expecting] Mary was expecting that John would come (20) Mary wa John ga kuru koto o kitaisite-ita (21) Mary wa John ga kuru to o kitaisite-ita There seems to be some subtle difference in meaning among these sentences. Sentence... [(21)] most likely means that the expectation did not come true. This seems to be related to the nonpresuppositional nature of to. Between... [(20) and (19)], it seems to be the case that the latter represents a stronger conviction on the part of the subject that John would come. This might be due, again, to the fact that no represents a concrete action, state, or event directly perceived by any of five (or six) senses, while koto represents a more abstract concept (Kuno 1973.221). This scale of ABSTRACT CONCRETE is used by Kuno to explain the contrasting acceptabilities of these sentences (Kuno 1973.222): (22) (a) Watakusi wa Columbus ga Amerika o [I America hakkensita no o siranakat-ta discovered know.not.past] I did not know (the fact) that Columbus discovered America 6 Jack Wiedrick comments about (18c) that This sentence is indeed kind of marginal, but could be read as: I learned (by instruction) that Japanese was difficult (but later that turned out not to be quite true). It would be better with a verb like osierareta was taught.

60 THE SEMANTICS OF SYNTACTIC COMPLEXITY (b) Watakusi wa Columbus ga Amerika o [I America hakkensita koto o siranakatta discovered know.not-past] I did not know (the fact) that Columbus discovered America (23) (a) *Watakusi wa kuzira ga honyuu-doobutu de aru [I whale mammal no o siranakat-ta know.not-past] (b) Watakusi wa kuzira ga honyuu-doobutu de aru [I whale mammal koto o siranakat-ta know.not-past] I did not know the fact that a whale is a mammal Sentence... [(22)] can be an abstract concept or a concrete event, but... [(23)] cannot be a concrete event. Therefore, it is expected that nominalizing... [(23a)] with no would result in an awkward clause (Kuno 1973.21-22).7 We may see in the stronger conviction of (19) expressed by no the maximum degree of integration between two events imposed by Mary s expectation (conviction?), whereas in (20) the bond is lessened and it is merely a matter of fact. The contrast between (22a) and (23a) is more elusive... and the more interesting. It may be that not knowing (but implied subsequent discovery) is more compatible... integrateable... with a historical occurrence, whereas supposed eternal truths (non-occurrences) such as whales being mammals are more remote and disconnected from John s learning... and hence not compatible with no. It does not seem unreasonable to apply Kuroda s and Tsubomoto s scheme of integration to these examples and to 7 Jack Wiedrick opbserves of (23a): I don t see anything wrong with this sentence. I m quite sure that any number of native speakers would accept it (with perhaps a caveat that the koto version is better in a neutral academic context). But it is especially acceptable in a context where you are thinking about the behaviour of whales, and perhaps surprised to discover that they suckle their young, you say (23a), meaning: I didn t know whales were mammals ( i.e. do mammal things like suckle their young, etc.)! In this sense, being a mammal is seen as a kind of lifestyle, rather than a classification, and (23a) seems quite appropriate.

JAPANESE 61 those in (24): (24) (a) John wa Mary ga sin-de simat-ta no o [ dying ended.up.with kii-ta hear-past] (b) John wa Mary ga sin-de simat-ta koto o [ dying ended.up.with kii-ta hear-past] John heard about the fact that Mary had died (c) John wa Mary ga sin-de simat-ta to kii-ta [ dying ended.up.with hear-past] John heard that Mary had died (Mary might or might not have died) Sentences (24b) and (24c) are from Kuno (1973.216), but we may expect that (24a) is also possible (kiku is a verb of perception) and that it would mean John heard Mary die.8 Cp. (10a). The relations that have emerged to this point suggest that no, koto, and to are related as in Figure 1. The three marks of complexity appear to form the CLASSIFICATION PRIMARY IDENTIFICATION SUPPLEMENTARY IDENTIFICATION MORE DISTINCT & NOT SEQUENCED MORE DISTINCT & CONSEQUENCED no koto to Figure 1: Japanese complexity. 8 Jack Wiedrick comments on (24a): This sentence could NEVER mean John heard Mary die. For that, you would have to use the verb kikoeru be audible : John niwa Mary ga sinde simatta no ga kikoeta. J. NI WA M. GA dying ended.up NO GA was.audible 'John heard Mary die.' (It s a strange sentence though; it kind of implies that he only heard her last dying gasp, and not anything before that [because of simatta].) The verb kiku hear, on the other hand, has more the sense of heard (because you asked, or through the grapevine), so (24a) means more-or-less the same thing as (24b), but there is a feeling that he might have heard all the gory details with no, whereas with koto it could only have been the statement Mary is dead (or the like).

62 THE SEMANTICS OF SYNTACTIC COMPLEXITY left most end of the continuum and to form it in a way that seems to ignore thelanguage distinctions of adjective, restrictive relative clause, nonrestrictive relative clause, complementiser, etc. 2.3 Adverbial clauses: to Adverbial clauses are also frequently marked by a to (Kuno 1973.184, 188 & 205): (25) John wa,boosi o nugu to, Mary ni aisatusi-ta [ hat take.off greet-past] John, upon taking off his hat, greeted Mary (26) Gakkoo ni iku to Mary ga ki-te i-ta [school go come-gerund Copula-Past] When I got to school, Mary had already been there (27) Natu ni naru to New York ni ikimasu [summer become go] When summer comes, I go to New York (28) Ie o deru to, ame ga hutte-ita [house leave rain falling-was] On my leaving home, it was raining... there is no logical connection between the speaker s leaving and it s raining... At the same time, S 1 and S 2 cannot be two independent events: S 1 to S 2 is acceptable only when the events represented by S 1 and S 2 are such that they are amenable to the paraphrase Upon S 1 s happening (or while S 1 was happening), what do you think happened? S 2 did.... Because of the what do you think happened next? connotation, the S 1 to S 2 construction brings the flavor of suspense or surprise (Kuno 1973.188-89). The temporal sequencing of to-clauses is necessary; it is the apparent overlapping of raining and pouring which damages (29b) (Kuno 1973.188): (29) (a) Ame ga huru to, nukarumi da [rain fall muddy is] When it rains, it becomes muddy

JAPANESE 63 (b)?ame ga huru to, dosyaburi da [rain fall squall is] Whenever it rains, it pours And (Kuno 1973.201): (30) John wa uwagi o nugu to, hangaa ni kakemasi-ta ka [ jacket take.off hanger hang-past Q] Taking off his jacket, did John hang it on a hanger? Sentence (30)... has only one interpretation, namely,... The speaker takes it for granted that John took off his jacket, and [(30)] cannot also be understood as questioning whether it is true that John took off his jacket and hung it on a hanger... In... [(30)] the speaker takes it for granted that John took his jacket and then questions whether he hung it on a hanger or not (Kuno 1973.201). The possibility of surprise and the absence of logical connection in (28) and the distinction between the events in (30) so that only one of them is questioned with ka, are consonant with the greater looseness associated with to, which is therefore appropriate to the expression of adverbial relations of when, on V-ing, etc. 3. Further elaboration of the continuum: te The te-gerundive will also be used to relate complex expressions (Kuno 1973.195-96): (31) Watasi wa anata ni it-te hosii [I you go-gerund want] I want you to go (32) John wa uwagi o nui-de hangaa ni kake-ta [ jacket take.off-gerund hanger hang-past] John took off his jacket and put it on a hanger (33) John wa hikoozyoo ni it-te, nimotu kensa o [ airport go-gerund luggage inspection uke-ta undergo-past] John went to the airport and underwent luggage inspection

64 THE SEMANTICS OF SYNTACTIC COMPLEXITY The V 1 -te V 2 pattern cannot be used when two simultaneous actions or states are involved... the two actions involved must be either both self-controllable or both non-self-controllable... This is why... [(33)] has, as one of its readings, the connotation that the speaker went to the airport in order to [Emphasis mine, PWD] undergo the luggage inspection. (Kuno 1973.196-97) The difference between to and te can be seen in the contrast between (30) and (34) (Kuno 1973.201): (34) John wa uwagi o nui-de, hangaa ni [ jacket take.off-gerund hanger kakemasi-ta ka hang-past Q] Did John take off his jacket and hang it on a hanger? In... [(34)] whether it was true that John took off his jacket and hung it on a hanger is being questioned. In... [another sense of (34)] the speaker takes for granted that John took off his jacket and then questions whether he hung it on a hanger or not [as in (30)]. (Kuno 1973.201) In contrast with (30), in which only the hanging of the jacket is questioned, both components of (34) are questioned. Like (30), sentence (34) questions the hanging of the jacket took place. The difference lies in how the taking off of the jacket relates its being hung up. In (34), the two actions appear sufficiently distinct that each is subject to independent questioning, while in (30), the activity is more complex. There is one event... hanging up a jacket... which is made more involved by the presence of more than one aspect to the achievement of the task, namely that the jacket has been taken off. But the removal of the jacket in (30) does not stand as an event in its own right. Hence, in (30) it is not questionable. Figure 1 is now expanded to Figure 2. CLASSIFICATION & PRIMARY IDENTIFICATION & SUPPLEMENTARY IDENTIFICATION MORE DISTINCT & NOT SEQUENCED MORE DISTINCT & CONSEQUENCED STILL MORE DISTINCT & CONSEQUENCED no koto to te Figure 2: Japanese complexity.

JAPANESE 65 3.1 Tara A more precise segmentation is possible among the when/if-clauses. Compare (27) with (35) (Kuno 1973.184): (27) Natu ni naru to New York ni ikimasu [summer become go] When summer comes, I go to New York (35) Natu ni nat-tara New York ni ikimasu [summer become go] When summer comes, I go to New York... [(27)] means that the speaker goes to New York every summer, while... [(35)] says that he intends to go to New York the coming summer. In other words,... [(27)], with to, represents habitual actions, while... [(35)], with tara, represents a single future action [of going. Emphases mine, PWD] (Kuno 1973.185). The two events expressed in (27) appear more connected, than the two in (35). Thus, (36a) is acceptable because there is in returning home no necessary connection between the two; but when consequence is imposed as in (36b) by the shared performer and the events of returning home and cooking, then tara fails (Kuno 1973.180-81): (36) (a) Ie ni kaet-tara, tegami ga ki-te ita [home return- letter come-gerund had] When I returned home, the letter had arrived (b) *Ie ni kaet-tara, gohan o tukut-ta [home return- meal prepare-past] When I returned home, I cooked dinner This indicates that tara-clauses are less tightly bound than to-clauses, which require some greater sense of connection. Compare also (Kuno 1973.180): (37) (a) John ga uwagi o nui-dara, Mary ga hangaa ni [ jacket take.off- hanger kake-ta hang-past] When John took off his jacket, Mary hung it on a hanger

66 THE SEMANTICS OF SYNTACTIC COMPLEXITY (b) *John ga uwagi o nui-dara, hangaa ni kake-ta [ jacket take.off- hanger hang-past] When John took off his jacket, he hung it on a hanger... in... [(37b)] the Agent of S 1 and that of S 2 are identical (namely, John), and since the action of taking off one s jacket and that of hanging it on a hanger are both self-controllable, it is considered that there was a self-controllable time sequence between S 1 and S 2. Hence, the sentence is ungrammatical. One the other hand, in... [(37 a)] the agent of sentence S 1 and the agent of S 2 are not identical, and although each action must have been self-controllable on the part of the agent, the time sequence does not imply that Mary hung John s jacket after waiting for him to take it off. It simply means that Mary s hanging John s jacket on a hanger followed [Emphases mine, PWD] John s taking it off... S 2 represents an unexpected or surprising event (Kuno 1973.170-71). 9 Both (36b) and (37b) contain clauses that bear some sense of control... or consequence... between them, which conflicts with tara, which signals precisely the absence of such consequence. The contrast between the unquestioned version of (30) with to (Kuno 1973.200) and (32) with te and (37) with tara is instructive: (30) John wa uwagi o nugu to, hangaa ni kakemasi-ta [ jacket take.off hanger hang-past] Taking off his jacket, John hung it on a hanger (32) John wa uwagi o nui-de hangaa ni kake-ta [ jacket take.off-gerund hanger hang-past] John took off his jacket and put it on a hanger 9 Jack Wiedrick comments as follows about (37b): It s true that with John ga it can t really be John who hung up his own coat (it s simply unspecified, but you assume it probably wasn t him, or they would ve said it differently). But the sentence is fine in the reading: After John took off his coat, (I, the doorman, etc.) hung it on a hanger. To have it be John, you could use -te instead (as Kuno suggests), or change it to John wa, in which case the sense would be: John, after taking off his coat, hung it on a hanger. Using -tara makes it seem like taking off the coat was an obstacle to be overcome (maybe it took a while), and that hanging something (anything) on the hanger was his intention from the start, i.e. it was necessary that he remove his coat before he could hang it on the hanger. It would seem natural in a scene where he enters a house, only to have the host confront him immediately and proffer a hanger, at which he would feel obligated to remove his coat in haste so as not to rudely keep the anxious host waiting.

JAPANESE 67 (37) (a) John ga uwagi o nui-dara, Mary ga hangaa ni [ jacket take.off- hanger kake-ta hang-past] When John took off his jacket, Mary hung it on a hanger (b) *John ga uwagi o nui-dara, hangaa ni kake-ta [ jacket take.off- hanger hang-past] When John took off his jacket, he hung it on a hanger In (30), the removal of the jacket is a component of its being hung up. In (32), the two events are more distinct but connected so that the second is a consequence of the first. And in (37), the two are still more distinct in that the second is not a consequence of the first. Recall that the second constitutes an unexpected or surprising event in the context of the first. In terms of Figure 2, tara will be... presently... the rightmost. 3.2 Nara The last clause type of this sort is introduced by nara. We may set it in contrast with to and tara from (27) and (35) (Kuno 1973.170): (27) Natu ni naru to New York ni ikimasu [summer become go] When summer comes, I go to New York (35) Natu ni nat-tara New York ni ikimasu [summer become go] When summer comes, I go to New York (38) (a) *Natu ni naru nara, New York ni ikimasu [summer become if go] If summer is to come, I will go to New York (b) Ima sugu naru nara, New York ni ikimasu [now immediately become go] If summer is to come right now, I will go to New York Sentence (27) represents the complex so that there is a necessary connection between summer coming and going to New York (the habitual trip), while in

68 THE SEMANTICS OF SYNTACTIC COMPLEXITY (35) there is no necessary connection, and it can be a one-time occurrence. But nara does not allow the comparable expression. Cf. (38a).10 The reason is that with nara S 1 represents a state or an action that the hearer can assert [Emphasis mine, PWD] (Kuno 1973.170). Another aspect of nara is illustrated by (39) (Kuno 1973.170-71): (39) (a) John ga kuru nara, kaerimasu [ come if return] If John comes, I will leave (resolution) (b) % Johnga kuru nara, Mary ga kaerimasu [ come if return] If John comes, Mary will leave (resolution)... [(39b)] would be acceptable if one imposes upon Mary ga kaerimasu Mary will leave the causative interpretation I will make Mary leave (Kuno 1973.171). The S 2 of S 1 nara S 2 must represent the speaker s evaluation, will, resolution, request, or order (Kuno 1973.170). In general, the first clause marked by nara cannot have an action or event in S 2 whose realization depends upon that of a future action or event represented by S 1 (Kuno 1973.173). Thus, (40) (a) Tabako o nomu nara, seiramu ga ii [tabacco smoke Salem good] If you are going to smoke (as you say you are), Salem is best (b) *Tabako o nomu nara, yamerarenaku naru desyoo [tabacco smoke addicted become] If you are going to smoke (as you say you are), you will be addicted to it In... [(40b)],... the hearer will become unable to stop smoking... only after 10 Jack Wuedrick comments as follows on (38a): This is possible if the meaning is something like: If (the plans are) pushed ahead to summer, I ll go to New York. Or, if you live in a place which never experiences a real summer (say, Antarctica), and yet you heard a forecast that summer would indeed come that year (but you re not holding your breath), then it s also a possible sentence in Kuno s intended reading.

JAPANESE 69 he has started smoking (Kuno 1973.173), and this imputed connection is what disables the sentence.11 The connection between events depicted by nara seems even more loose than that of dara. There is a sense of assertion that is absent from the preceding.12 4. Coördination Coördination is signalled by to (and ni and ya) for nouns and by si for clauses (Kuno 1973.114 & 202): (41) John to Mary ga kekkonsi-ta [ marry-past] John and Mary became man and wife John got married and Mary got married 11 Jack Wiedrick disputes Kuno s comments as follows: Kuno s judgement here is simply not true. There s nothing wrong with this sentence as it stands (except for his choice of verb, which marks him as an old fogy!). It would be ok as a conditional (e.g. listing-pros-and-cons-type advice), but as a warning or censure, though, it is particularly appropriate, i.e.: If you re dead set on smoking, then I guess you can expect to get addicted. Kuno assumes that the only reading is one where the smoking has not yet taken place, but the sentence actually makes more sense when said to a person who is already a smoker (and who has most likely complained about not being able to quit). The sense is kind of: Don t come bitching to me, or You made your bed, now lie in it. Also, what Kuno (1973.170) says is true about B not allowing (unmodified) verbs in -ta in A nara B, but this is not because of any so-called constraint that [B] must represent the speaker s evaluation, will, resolution, resolution, or order. That s not true at all; I can come up with dozens of perfectly fine sentences that break the constraint. Consider: nisinhoonara, 5 wa 101 da base.2 NARA 5 WA101 is If (we count in) binary, (then) 5 is 101 The statement is neither my will nor my evaluation; it is simply a mathematical fact. It s true that nara is often used in situations where the speaker is making a statement of will or intent, but this is not the ONLY usage possible. As you can see, nara can also be used when stating a logical or inevitable consequence of something in the protasis. What makes the -ta troublesome in the apodosis is simply this fact of the timeless character of nara. However, if you push the apodosis into irrealis (for lack of a better word), then it becomes ok: Taroo ga kita nara, Mary wa naita daroo T. GA came NARA M. WA cried probably If Taroo came, (then) I ll bet that Mary cried 12 Kuno (1973.168) writes that It is usually said that this pattern has a strong degree of assertion about the statement represented by S 1.

70 THE SEMANTICS OF SYNTACTIC COMPLEXITY (42) John wa e o kaku si haiku o tukuru [ picture paint poem compose] John paints pictures and composes poems (43) *John wa e o si haiku o tukuru ka [ picture poem compose Q] Sentence... [(42)] is an example of pure coordination by si and, which is used to juxtapose two events without saying which event took place first. It seems that the command power of ka cannot be extended to the left of si. Example... [(43)] is ungrammatical because the intended function of ka is to make the whole sentence a question, while si prevents the command power of ka from being extended to the left of it. Thus,... [(43)] ends up with the juxtaposition of a declarative clause and an interrogative clause, which is an ungrammatical combination in Japanese, as well as in English and many other languages (Kuno 1973.203). The coördinate clauses with si force the stringing together of the clauses in a way that appears to mirror the looser semantic relation of GENERAL RELE- VANCE. The coördinate clause contrasts with the temporal to-clause in this way (Kuno 1973.205-06): (44) (a) John wa,boosi o nugu to, Mary ni aisatusi-ta [ hat take.off greet-past] John, upon taking off his hat, greeted Mary (b)?john wa,mary ni, boosi o nugu to, aisatusi-ta [ hat take.off greet-past] John, upon taking off his hat, greeted Mary (45) (a) John wa, e o kaita si, haiku o tukut-ta [ picture paint poem compose-past] John painted pictures and composed poems (b) *John wa, haiku o, e o kaita si, tukut-ta... the grammaticality (in spite of its awkwardness) of... [(44b)] indicates that the V 1 to is still subordinate to V 2. If V 1 and V 2 formed a coordinate structure V 1 and V 2, no element of V 2 could be preposed to the left of V 1... The awkwardness of... [(44b)] indicates that V 1 to V 2 has a status somewhere between the subordinate phrase main phrase construction and the coordinating construction (probably closer to the former than to the latter (Kuno 1973.206).

JAPANESE 71 5. Conclusion Figure 3 summarizes the patterns of Japanese. One of the prominent con- CLASSIFICATION & PRIMARY IDENTIFICATION & SUPPLEMENTARY IDENTIFICATION MORE DISTINCT & NOT SEQUENCED MORE DISTINCT & CONSEQUENCED STILL MORE DISTINCT & CONSEQUENCED STILL MORE DISTINCT & NOT CONSEQUENCED ASSERTED & NOT CONSEQUENCED CONJOINED no koto to te tara nara si Figure 3: Japanese complexity. clusions is that the organization of Japanese is not in terms of the classifications which we impose from the perspective of linguistics. Examination of the Japanese examples shows that English notions such as when, if, in order to, etc. (and the linguistic concepts that they inspire) are scattered across the various ways of expressing complexity in Japanese. The functional categories of CLASSIFICATION, PRIMARY IDENTIFICATION, SUPPLEMENTARY IDENTIFICATION, SUPPLEMENTARY REMARK, etc. are accidental in the organization of Japanese, yet Japanese is very sensible... in its own terms. The principle appears to be one in which two occurrences, first seen as merged as two suns locked into one orbit, gradually draw apart until they occupy their own orbits and accept their separate identities. The content manipulated and organized by the Japanese clausal postpositions is visually presented in Figure 4. The principle which guides the visualization is that of merger, encapsulation, and identity at the left of the Figure... a condition that is then gradually attenuated across the complex expressions, until at the right, there is a complete separation so that the content of neither PROPOSITION is conceived as belonging to some larger conceptualino koto to te tara nara si Figure 4: Semantic organization of complexity in Japanese.

72 THE SEMANTICS OF SYNTACTIC COMPLEXITY zation of what happened. One of the constants of no is the Relevancy Condition; and no13 and koto both mark the contents of their PROPOSITIONS, not only as unquestionable, but also as not contingent. Their content is so in the way that calling a book a book is so. It just is. With the postposition mark to, the clause is more contingent14, and although the Relevancy Condition appears to be no longer in force, but a connection of consequence (something like and then ) continues to relate the contents of the two PROPOSITIONS. The portion marked by to cannot be questioned; and this condition contrasts to with te, which does permit the content of its clause to be queried. The postposition tara loosens the intertwining of the contents of two PROPOSITIONS still more in that there is now no necessary connection between the two contents; they are juxtaposed pieces of some larger occurrence.15 The use of nara differs from tara in that its PROPOSITION has a stronger degree of independence from the whole and some component of assertion.16 Finally, si juxtaposes events without relating them as to which came first. They are now each their own event. This condition is expressed in Figure 4 by the removal of the two events from their encapsulating ring. It is important to keep in mind that the shapes in Figure 4... and in the Figures below... are arbitrary. They are only an attempt to make the intuitive interpretations of the patterns of Japanese more concrete. With this reservation, Figure 5: Semantic organization of complexity in Bella Coola. those shapes may then be distributed in Bella Coola, Farsi, and in Warlpiri as in Figures 5, 6, and 7. The results may in certain cases seem artificial; and if 13 Recall Kuroda s association of simultaneity to no, and Tsubomoto s claim that events linked with no are as if they were one event. 14 Kuno asserts that there is no logical connection between events linked with to, and that to brings a flavor of suspense or surprise. 15 Kuno s comment on tara was that it signalled a single future action that simply... followed. 16 It is an expression of the speakers evaluation, will, resolution, request, or order (Kuno).

JAPANESE 73 Figure 6: Semantic organization of complexity in Farsi. Figure 7: Semantic organization of complexity in Warlpiri.17 this is so, it is because the language in question fails to exploit the organization present in Japanese (both in number and configuration). 17 An alternative to Figure 7 might identify the relation marked by the infinitival construction (cf. [i]) with that of the adjectival (cf. [ii]. Simpson 1988.211), thus merging the two distinctions on the left of Figure 7: (i) [I Aux grief-inchoative-nonpast you-dat sick lie-infinitive-comp-dat] I feel sorry for you while you are lying sick (ii) [Japanangka-Erg-Conative shoot-past kangaroo-dative rock-loc-dative big-loc-dative] Japanangka shot at the kangaroo on the big rock There are several reasons for making this equation. First, the infinitival construction expresses a relation between the two PROPOSITIONS that is tighter than that expressed by / (the adjoined relative markers) and certainly more integrated (Cp. Hale s [1976.82] ongoing or in effect at the time and the if induced by irrealis combined with katji in Warlpiri [cf. sentence (50) in Chapter 3].) than that expressed by (the purposive clauses) or (conjoined clauses). This semantic relation between the infinitival clause and the other clause forms places the infinitival construction adjacent to (or at the same place on the scale if the proposed identification is accepted) as the adjectival expressions. Second, the infinitival clauses are marked by a nominaliser, which creates forms which may then take case marking. Compare the occurrence of the dative in the infinitival form in (i) with the adjectival form in (ii). Third, both the infinitival form and the adjectival form may be interlaced within the remainder. Infinitive clauses... have the ability to appear within the main clause and to permute with other constituents of it (Hale 1976.94). The clauses related by /,, and may not show this formal interconnection (Nash 1986.240).

74 THE SEMANTICS OF SYNTACTIC COMPLEXITY