1. Introduction
External possession constructions (EPCs) across languages have been examined from two opposing views:
base-generation versus raising. In such constructions, the possessor is a clausal element external to the posses- sum phrase, but undergoes interpretation as the possessor of the head noun (i.e., the possessee) within a posses- sum phrase. The logical question is how the possessor-possessee relation is established. The base-generation analysis holds that the possessor is base-generated external to the possessum phrase, and is associated with the possessee via, for instance, some type of predication (Heykock 1993, Ogawa 2001), recursive VP structure (Tomioka and Sim 2007), or control (Doron and Heykock 1999, 2009). On the other hand, the movement analy- sis posits that the possessor is raised from within the possessum phrase, hence the possessor-possessee relation (e.g., Ura 1996, Landau 1999). The following EPC in Korean is one such example subject to the two kinds of syntactic analyses (see Yoon (2015) for a recent review of this type of double accusative construction in Korean):
(1) a. Chelswu-ka Sunhee-euy son-u cap-ass-ta.
Chelswu-NOM Sunhee-GEN hand-ACC grab-PAST-DECL
‘Chelswu grabbed Sunhee by the hand.’
Notes on a New Type of External Possession Construction in Japanese*
Seiki AYANO
Abstract
External possession constructions (EPCs) found crosslinguistically are under two opposing analyses: base-genera- tion versus movement. One argument against the movement analysis is that the affectedness condition applies to EPCs; typically, inalienable possessive relations are required for EPCs. This paper shows that in double-goal ditran- sitive constructions similar to those discussed in Miyagawa and Tsujioka (2004) are actually EPCs, and that in these constructions, inalienable possessive relations are not a requirement for the availability of EPCs. While the move- ment analysis of this type of EPC can be empirically supported, there is counterevidence against such analysis.
These contradicting facts can be best explained by Hornstein’s (1999) proposal on control by movement. Finally, this paper discusses the syntactic structure of EPCs in question.
* Research for this paper is supported in part by a Grant-in-Aid for Scientific Research (C) from the Japan Society for the Promotion of Science (#19K00548). An earlier version of this paper was presented at the 19th Seoul International Conference on Generative Grammar held at Seoul National University. I would like to thank the audience for their comments. The usual disclaimers apply.
b. Chelswu-ka Sunhee-lul son-ul cap-ass-ta.
Chelswu-NOM Sunhee-ACC hand-ACC grab-PAST-DECL
‘Chelswu grabbed Sunhee by the hand.’ (Tomioka and Sim 2007)
In (1), there is an inalienable possessive relation between the possessor and the possessee, and the possessor can appear either in the genitive case or in the accusative case, while in the absence of such relation, double accusa- tive constructions are unavailable, as illustrated in (2b).
(2) a. John-i Mary-uy kon-lul cap-ass-ta.
John-NOM Mary-GEN ball-ACC catch-PAST-DECL
‘John caught Mary’s ball.’
b. *John-i Mary-lul kon-lul cap-ass-ta.
John-NOM Mary-ACC ball-ACC catch-PAST-DECL
‘John caught Mary’s ball.’ (Ura 1996)
Under the base-generation analysis, Yoon (1990) and Tomioka and Sim (2007) argue that the unavailability of double accusative constructions is due to the affectedness condition, where the possessor as well as the posses- see needs to be affected in order for double-accusative constructions to be licensed. That is, the possessor Sunhee-lul ‘Sunhee-ACC’ as well as the possessee son-ul ‘hand-ACC’ in (1b) is affected by the event of Chelswu’s grabbing, whereas the possessor Mary-lul ‘Mary-ACC’ in (2b) is not. Thus, the ungrammaticality of (2b) can be accounted for.
In contrast to EPCs such as (1b), in which inalienability is a condition on the availability of EPCs, Deal (2013) has shown that it is not the case in Nez Perce. Consider (3), in which the possessor is ha-haacwal-na ‘PL-boy- OBJ’ and the possessee is lawtiwaa ‘friend.NOM’; the possessive relation between the two is an alienable one.
(3) Haama-pim hi-nees-wewkuny-e’ny-∅-e ha-haacwal-na lawtiwaa.
man-ERG 3SUBJ-O.PL-meet-μ-P-REM.PAST PL-boy-OBJ friend.NOM
‘The man met the boys’ friend.’ (Deal 2013)
Note that in (3), the possessor ha-haacwal-na ‘PL-boy-OBJ’ is a clausal element, in that it is in agreement with the verb in number and is suffixed by the object (accusative) marker. Deal (2013) provides evidence for the movement analysis and discusses the syntax of this type of EPC.
In Ayano (2017), I have argued that double-goal ditransitive constructions à la Miyagawa and Tsujioka (2004), such as (4) given below, represent EPCs that do not require an inalienable possessive relation between the possessor and the possessee: 1
(4) Taroo-ga Hanako-ni (kossori) zitaku-ni hanataba-o todoke-ta.
Taroo-NOM Hanako-NI secretly home-NI bouquet-ACC deliver-PAST
‘Taroo (secretly) delivered Hanako a bouquet to her home.’
The possessor Hanako can appear in the genitive case, and as the ungrammaticality of (5) with the presence of the intervening adverb kossori ‘secretly’ indicates, the genitive possessor remains within the locative possessum phrase.
(5) Taroo-ga Hanako-no (*kossori) zitaku-ni hanataba-o okut-ta.
Taroo-NOM Hanako-GEN secretly home-NI bouquet-ACC send-PAST
‘Taroo (secretly) sent Hanako a bouquet to her home.’
This paper explores the syntax of double-goal ditransitive constructions such as (4). It is organized as follows.
Section 2 will provide empirical support for the movement analysis based on honorification and scope. Section 3 will begin with counterevidence against the movement analysis, and argue that Hornstein’s movement analysis of control can account for the seemingly contradictory facts. Finally, Section 4 concludes this paper.
2. The syntax of double-goal ditransitive constructions 2.1. Evidence for the movement analysis
At the outset of my discussion on the syntax of double-goal ditransitive constructions in Japanese, let me exam- ine the syntactic properties of the two -ni phrases in (4) above. The following contrast with respect to numer- al-quantifier floating between (6a) and (6b) suggests the categorical properties of the two types of -ni:
(6) a. Sobaya-ga zyoyuu-ni huta-ri gakuya-ni demae-o todoke-ta.
noodle.shop-NOM actress-DAT two-CL dressing.room-to meal-ACC deliver-PAST
‘The noodle shop delivered two actresses meals to the theater’s dressing room.’
b. *Sobaya-ga zyoyuu-ni gakuya-ni itutu-tu demae-o todoke-ta.
noodle.shop-NOM actress-DAT dressing.room-to five-CL meal-ACC deliver-PAST
‘The noodle shop delivered the actresses meals to five (different) dressing rooms.’
1 (4) is a double-goal ditransitive construction similar to the ones discussed in Miyagawa and Tsujioka (2004), but is different in one respect. (i) is taken from their paper:
(i) Taroo-ga Hanako-ni Tokyo-ni nimotu-o okut-ta.
Taroo-NOM Hanako-NI Tokyo-NI package-ACC send-PAST ʻTaro sent a package to Hanako to Tokyo.ʼ
Notice that the possessor Hanako-ni ʻHanako-NIʼ in (i) cannot appear in the genitive case, as shown in (ii):
(ii) *Taroo-ga Hanako-no Tokyo-ni nimotu-o okut-ta.
Taroo-NOM Hanako-GEN Tokyo-NI package-ACC send-PAST ʻTaro sent a package to Hanako to Tokyo.ʼ
This paper will focus on the syntactic structure of the type of double-ditransitive construction illustrated by (4), and argue that it is derived by possessor-raising from within the possessum phrase.
In (6a), the possessor suffixed by -ni permits numeral-quantifier floating that is a hallmark of identifying case-markers. In contrast, -ni suffixed to the goal in (6b) does not allow numeral-quantifier floating, which shows that it is a postposition rather than a case-marker. Thus, the possessor is in the dative case. In what follows I will show that the construction in question involves possessor-raising.
The first piece of evidence is based on honorification licensing. Kishimoto (2013) claims that EPCs in Japa- nese involving complex idiomatic predicates such as ki-ni sawaru ‘feel annoyed’ and kuti-ni awu ‘favor’ can be explained by possessor-raising. Consider (7)
(7) a. Sono-koto-ga Ken-no (*mezurasiku) ki-ni sawat-ta (rasii) that-fact-NOM Ken-GEN unusually mind-LOC harm-PAST seem
‘(It seems that) Ken did not (unusually) like that fact.’
b. Ken-ni sono-koto-ga (mezurasiku) ki-ni sawat-ta (rasii) Ken-DAT that-fact-NOM unusually mind-LOC harm-PAST seem
‘(It seems that) Ken did not (unusually) like that fact.’
In (7a), the possessor Ken ‘Ken’ is suffixed by the genitive case-marker, and the adverb mezurasiku ‘unusually’
cannot intervene between the possessor and the possessee, which suggests that the possessor is within the pos- sessum phrase. On the other hand, in (7b), the possessor is in the dative case, and the intervention of the adverb mezurasiku ‘unusually’ between the possessor and the possessee does not result in ungrammaticality.
Kishimoto (2013) argues that honorification facts support the movement analysis of (7b). (8) demonstrates that honorification can be licensed within the possessive phrase; the honorific marker o- can target the genitive possessor (Kishimoto 2013).
(8) Ito-sensee-no o-kuti/o-me
Ito-teacher-GEN HON-mouth/HON-eye
‘Prof. Ito’s mouth/eye’
On the basis of (8), Kishimoto (2013) shows that this honorification licensing within the possessive phrase can be extended to EPCs such as (7b), which serves as evidence for possessor-raising. Consider (9):
(9) a. Sono-koto-ga Ito-sensee-no (*mezurasiku) o-ki-ni sawat-ta (rasii) that-fact-NOM Ito-teacher-GEN unusually HON-mind-LOC harm-PAST seem
‘(It seems that) Prof. Ito did not (unusually) like that fact.’
b. Ito-sensee-ni sono-koto-ga (mezurasiku) o-ki-ni sawat-ta (rasii) Ito-teacher-DAT that-fact-NOM unusually HON-mind-LOC harm-PAST seem
‘(It seems that) Prof. Ito did not (unusually) like that fact.’
In both (9a) and (9b), honorification is licensed on the respective possessors; the genitive possessor in (9a) and the dative possessor in (9b) are legitimate targets of the honorific marker o- ‘HON.’ Note that (10) with the tran- sitive idiom kuti-ni dasita ‘spelled out’ is in sharp contrast to (9b). That is, the honorific marker on the possessee
o-kuti-ni ‘HON-mouth-LOC’ cannot target the nominative possessor Ito-sensee-ga ‘Prof. Ito-NOM.’ This is be- cause the nominative possessor is base-generated outside the possessum locative kuti-ni ‘to mouth’ (Kishimoto 2013).
(10) Ito-sensee-ga human-o (*o-)kuti-ni dasi-ta.
Ito-teacher-NOM complaint-ACC HON-mouth-LOC put.out-PAST
‘Prof. Ito made a complaint.’
Applying this honorification diagnostic to the double-goal ditransitive construction at issue, we obtain the result in (11):
(11) Eri-ga osisyo-ni (kossori) go-zitaku-ni oseebo-o todoke-ta.
Eri-NOM teacher-DAT secretly HON-home-to year.end.present-ACC deliver-PAST
‘Eri (secretly) delivered a year-end-present to her teacher’s home.’
Note that the genitive possessor can also be a target of the honorific marker -go.
(12) Eri-ga osisyo-no (*kossori) go-zitaku-ni oseebo-o todoke-ta.
Eri-NOM teacher-GEN secretly HON-home year.end.present-ACC deliver-PAST
‘Eri (secretly) delivered a year-end-present to her teacher’s home.’
The second empirical support is based on scope. It has been observed that the nominative subject in three dif- ferent passive constructions in Japanese, namely, direct, indirect and possessive passives, interact with adverbs such as mai-asa ‘every morning’ (see Hasegawa 2009), and direct and possessive passives behave alike regard- ing scope ambiguity of their nominative subjects, as illustrated in (13):
(13) a. Direct passive
San-nin-no kodomo-ga sensee-ni mai-asa homer-are-tei-ru.
three-CL-GEN child-NOM teacher-NI every-morning praise-PASS-ASP-PRES
‘Three children are praised by the teacher every morning.’
✔ three children > every morning (Reading: same three children)
✔ every morning > three children (Reading: different sets of three children) b. Indirect passive
San-nin-no kodomo-ga oya-ni mai-asa neboos-are-tei-ru.
three-CL-GEN child-NOM parent-NI every-morning sleep.in-PASS-ASP-PRES
‘Three children have their parent(s) sleep in every morning.’
✔ three children > every morning (Reading: same three children)
✖every morning > three children (Reading: different sets of three children)
c. Possessive passive
San-nin-no kodomo-ga sensee-ni mai-asa atama-o nade-rare-tei-ru.
three-CL-GEN child-NOM teacher-NI every-morning head-ACC rub-PASS-ASP-PRES
‘Three children have their heads rubbed by the teacher every morning.’
✔three children > every morning (Reading: same three children)
✔every morning > three children (Reading: different sets of three children)
The scope ambiguity shown in the direct passive in (13a) and the possessive passive in (13c) indicates that their nominative subjects have raised from below the adverb mai-asa ‘every-morning.’ In contrast, the unambiguity, namely, the unavailability of the narrow reading of three children in the indirect passive, shows that the base po- sition of the nominative Experiencer subject is always higher than the adverb. On the basis of the above scope facts, provided that the dative possessor in double-goal ditransitive constructions involves movement, it should exhibit scope ambiguity in relation to the adverb placed between the dative possessor and the postpositional goal, while the postpositional goal placed below the adverb should not show ambiguity. The contrasts illustrated in (14) and (15) show that the prediction is correct:
(14) a. Borantia-ga zyuu-nin-no otosiyori-ni mai-ban otaku-ni syokuzi-o todoke-teir-u.
volunteer-NOM ten-CL-GEN elderly-DAT every-evening home-to meal-ACC deliver-PROG-PRES
‘Volunteers deliver ten elderly people meals to their homes every evening.’
✔ten elderly people > every evening (Reading: same ten elderly people)
✔every evening > ten elderly people (Reading: different sets of ten elderly people)
b. Borantia-ga otosiyori-ni mai-ban san-ken-no otaku-ni syokuzi-o todoke-teir-u.
volunteer-NOM elderly-DAT every-evening three-CL-GEN home-to meal-ACC deliver-PROG-PRES
‘Volunteers deliver elderly people meals to three homes every evening.’
✖three homes > every evening (Reading: same three homes)
✔every evening > three homes (Reading: different sets of three homes)
(15) a. Eri-ga san- nin-no kabuki-yakusya-ni mai-syuu gakuya-ni hanataba-o okut-teir-u.
Eri-NOM three-CL-GEN kabuki-actors-DAT every-week dressing.room-to bouquet-ACC send-PROG-PRES
‘Eri sends three kabuki actors bouquets to their office every week.’
✔three kabuki actors > every week (Reading: same three kabuki actors)
✔every week > three kabuki actors (Reading: different sets of three kabuki actors) b. Eri-ga kabuki-yakusya-ni mai-syuu ni-ken-no zitaku-ni hanataba-o okut-teir-u.
Eri-NOM kabuki-actors-NI every-week two-CL-GEN home-to bouquet-ACC send-PROG-PRES
‘Eri sends kabuki actors bouquets to two homes every week.’
✖two homes > every week (Reading: same two homes)
✔every week > two homes (Reading: different sets of two homes)
In (15a), the dative possessor, otosiyori-ni ‘to the elderly’ exhibits ambiguity with respect to the adverb, mai-ban
‘every evening,’ which can be taken to show that it has raised from below the adverb, while the lack of such am- biguity for the postpositional goal, otaku-ni ‘to home,’ indicates that it is always lower than the adverb. The same
contrast is observed between the dative possessor, kabuki-yakusya-ni ‘to kabuki actors,’ in (16a) and the postpo- sitional goal, zitaku-ni ‘to home,’ in (16b).
2.2. A problem for the movement analysis
I have provided evidence for the movement analysis of (4), but there is counterevidence against such analysis as well. Consider (4’), in which the genitive pronoun kanozyo-no ‘she-GEN’ can appear with the possessee zitaku
‘home.’
(4’) Taroo-ga Hanako-nik (kossori) kanozyo-nok zitaku-ni hanataba-o todoke-ta.
Taroo-NOM Hanako-DAT secretly she-GEN home-to bouquet-ACC deliver-PAST
‘Taroo (secretly) delivered Hanako a bouquet to her home.’
In fact, the possessee in Kishimoto’s complex idiomatic predicates can also be preceded by the genitive pro- noun kare-no ‘he-GEN,’ and the pronoun is coreferential with the possessor Ken-ni ‘Ken-DAT.’ Consider (7b’):
(7b’) Ken-nik sono-koto-ga (mezurasiku) kare-nok ki-ni sawat-ta (rasii) Ken-DAT that-fact-NOM unusually he-GEN mind-LOC harm-PAST seem
‘(It seems that) Ken did not (unusually) like that fact.’
Questions arise as to (i) how to understand the above fact in (4’) and (ii) how to reconcile this counterevidence with the evidence for the movement analysis provided in the previous section. In what follows, I will argue that the presence of the overt pronoun in (4’) is in parallel with that in finite control constructions, and that the move- ment analysis of control (Hornstein 1999) can account for the seemingly contradictory facts.
First, note that control constructions in English are limited to non-finite clauses:
(16) a. John tries to eat less red meat.
b. John promised Mary to eat less red meat.
However, crosslinguistically control constructions are not limited to the non-finite context. Landau (2004), for example, provides the following Hebrew example, in which the embedded clause is finite, as the agreement on the embedded verb shows, and the null subject of the embedded verb is obligatorily controlled by the matrix subject Gil.
(17) Gil hivtiax še yitna’heg yafe.
Gil promised that will-behave.3SG.M well
‘Gil promised to behave well.’
Note that the embedded subject can also be overt and coreferential with the matrix subject:
(18) Gili hivtiax še-hui yitna’heg yafe Gil promised that-he will-behave.3SG.M well
‘Gil promised to behave well.’
Other languages such as Korean and Japanese exhibit control in finite clauses, and in such context, overt pro- nouns/anaphors can appear in the embedded subject position and can undergo coreferential interpretation with the matrix subject (see Akuzawa 2017 for an extensive review).2
(19) a. John-i Bill-eykey ttena-keyss-ta-ko yaksokha-yess-ta.
John-NOM Bill-DAT leave-VOL-DC-C promise-PST-DC
‘Johni promised Bill that eci would leave.’
b. John-ii Bill-eykey kui/cakii-ka ttena-keyss-ta-ko yaksokha-yess-ta.
John-NOM Bill-DAT he/self-NOM leave-VOL-DC-C promise-PST-DC
‘John promised Bill that he/self would leave.’ (Lee 2009)
(20) a. Ken-wai kossori zibun-gai gitaa-o hik-u koto-o kokoromi-ta.
Ken-TOP secretly self-NOM guitar-ACC play-PRES C-ACC try-PAST
‘Ken tried that he played the guitar secretly.’
b. Ken-wai kossori zubun-gai gitaa-o hii-ta koto-o hanseesi-ta.
Ken-TOP secretly self-NOM guitar-ACC play-PAST C-ACC regret-PAST
‘Ken regretted that he played the guitar secretly.’
I claim that the possessum phrase containing the overt genitive pronoun in the double-goal ditransitive con- struction in (4’) is in parallel with the Korean and Japanese finite control clauses in (19b), (20a) and (20b). If we adopt the movement analysis of control (Hornstein 1999), then the empirical evidence for the movement analy- sis and the availability of overt genitive pronouns within the possessum phrase can be accounted for.3 Drawing on Hornstein’s movement analysis of control, the dative possessor in double-goal ditransitive constructions is considered (i) to be base-generated within the possessum phrase, where it is assigned a θ-role by the possessee, and (ii) to be assigned another θ-role by some functional head at the clausal level.
2 Regarding finite control clauses headed by -koto in Japanese, Watanabe (1996) and Uchibori (2000) argue that they are subjunctive and -koto is C.
3 See Funakoshi (2017) for his argument on backward control in EPCs. The grammaticality of the following example with a floated numeral quantifier (cf. Kikuchi 1994) can be accounted for by positing a null clausal element coreferential with the genitive possessor. This null clausal element is licensed by a null applicative head that denotes “inherent participation”
of the possessor in a given event.
(i) Hanako-wa kodomotati-no tume-o san-nin ki-ta.
Hanako-TOP children-GEN nail-ACC three-CL clip-PAST ʻHanako clipped three children’s nails.ʼ
2.3. The syntax of double-goal ditransitive constructions
I have argued in my earlier paper (Ayano 2017) that the landing site of the dative possessor is within vP, on the basis of (i) -bakari modification and (ii) indeterminate pronouns licensed by the Q-particle -mo.
First, the suffix -bakari ‘only’ can attach to V and DP (Kishimoto 2009, 2010). When it attaches to V, it can modify VP-internal elements only. Consider (21):
(21) Kodomo-ga manga-o yonde-bakari i-ru child-NOM comics-ACC reading-only be-PRES
‘The child is reading only the comics.’
In (21), the direct object manga-o ‘the comics-ACC’ can fall within the focus of -bakari ‘only,’ but not the sub- ject kodomo-ga ‘child-NOM.’ Thus, (21) is synonymous with (22b), not (22a).
(22) a. Kodomo-bakari-ga manga-o yonde-i-ru child-only-NOM comics-ACC reading-be-PRES
‘Only the child is reading the comic books.’
b. Kodomo-ga manga-bakari-o yonde-i-ru child-NOM comics-only-ACC reading-be-PRES
‘The child is reading only the comics.’
Applying this diagnostic to double-goal ditransitive constructions, we obtain the results in (23).
(23) Uti-no-ko-ga okiniiri-no guruupu-ni zumusyo-ni nigaoe-o todokete-bakari i-ru.
our-GEN-child-NOM favorite-GEN group-DAT office-to portrait-ACC deliver-bakari be-PRES
✖’Only our child sends portraits to the favorite group (of singers) to their office.’
✔’Our child sends only portraits to the favorite group (of singers) to their office.’
✔’Our child sends portraits only to the favorite group (of singers) to their office.’
✔’Our child sends portraits to the favorite group (of singers) only to their office.’
The dative possessor okiniiri-no guruup-ni ‘to the favorite group (of singers)’ is within the focus of -bakari at-
He assumes the movement analysis of control (Hornstein 1999) and the copy theory of movement (Chomsky 1995), and claims that the derivation of (i) proceeds in such a way that the higher copy of the possessor at the clausal level gets deleted, and the lower copy in the possessum phrase is pronounced. Funakoshi argues that which copy to be pronounced can be completely optional, as double-accusative constructions in Korean demonstrate (see example (1)). As regards the dative possessor in double-goal ditransitive constructions discussed here, the choice does not seem to be optional although it can alternate with the genitive possessor. As discussed towards the end of this paper, the dative possessor is licensed by an applicative head of a different kind from the one proposed by Funakoshi (2017), and it enters the derivation only when the dative possessor is present. It remains to be seen why genitive pronouns can be pronounced optionally as shown in (4’).
Regarding this issue, Takezawa (2016) argues that the presence of pronouns in finite control constructions involves pragmatics. The effect of the overt pronoun in (4’) may be explained in the same manner, but I leave it for further study.
tached to the ditransitive verb todokeru ‘deliver,’ which shows that the raised dative possessor remains within VP.
Second, negative polarity items such as dare-mo ‘anyone’ or nani-mo ‘anything’ must be licensed by negation:
(24) a. Dare-mo ohirugohan-o tabe-nakat-ta.
anyone-Q lunch-ACC eat-NEG-PAST
‘No one ate lunch.’
b. Taroo-ga nani-mo tabe-nakat-ta.
Taroo-NOM anything-Q eat-NEG-PAST
‘Taroo did not eat anything.’
The Q-particle attached to verbs can bind vP-internal elements such as the direct object, but not the subject that is outside vP (Kishimoto (2001).
(25) a. *Dare-ga ohirugohan-o tabe-mo si-nakat-ta.
anyone-Q lunch-ACC eat-Q do-NEG-PAST
‘No one did not eat lunch.’
b. Taroo-wa nani-o tabe-mo si-nakat-ta.
Taroo-TOP anything-ACC eat-Q do-NEG-PAST
‘Taroo did not eat anything.’
Similarly, vP-internal elements in double-goal ditransitive constructions should be bound by the Q-particle that attaches to verbs.
(26) a. *Dare-ga otosiyori-ni sisetu-ni yuuhan-o todoke-mo sitei-na-i.
anyone-NOM elderly-DAT nursing.home-to dinner-ACC deliver-Q doing-NEG-PRES
‘No one delivers the elderly dinner to the nursing home.’
b. Borantia-wa dare-ni sisetu-ni yuuhan-o todoke-mo sitei-na-i.
volunteer-TOP anyone-DAT nursing.home-to dinner-ACC deliver-Q doing-NEG-PRES
‘Volunteers do not deliver anyone dinner to the nursing home.’
c. Borantia-ga otosiyori-ni doko-ni yuuhan-o todoke-mo sitei-na-i.
anyone-NOM elderly-DAT anywhere-to dinner-ACC deliver-Q doing-NEG-PRES
‘Volunteers do not deliver the elderly dinner to anywhere.’
d. Borantia-wa otosiyori-ni sisetu-ni nani-o todoke-mo sitei-na-i.
anyone-TOP elderly-DAT nursing.home-to anything-ACC deliver-Q doing-NEG-PRES
‘Volunteers do not deliver the elderly anything to the nursing home.’
The grammaticality of (26b) shows that the dative possessor is considered to remain within vP.
Given that the dative possessor is within vP, the relevant derivation of (4) is illustrated in (27):
(27) [vP Poss-ni [?P Poss-ni … [PP Poss-ni ... [NP Poss-ni possessee]]-ni ] ]4
Kishimoto (2013) and Aoyagi (2015) claim that dative case of ditransitive arguments is licensed by v. Following their analysis, I assume that that the dative case on the possessor is licensed in [Spec, v]. Concerning θ-role as- signment (or θ-role checking (Hornstein 1999)), the possessee assigns a θ-role in the base position. The question is what assigns a θ-role to the dative possessor at the clausal level. In order to address this issue, let us consider the argument structure of verbs, which permits double-goal ditransitive constructions.
The verbs used in the double-goal ditransitive constructions discussed in this paper belong to verbs of transfer, as opposed to verbs of change of possession according to Kishimoto’s (2001) classification. Properties of -ni phrases selected by the two types of verbs can be demonstrated by two diagnostics: passivization and -ni/-e alter- nation. (28) and (29) exhibit a contrast between verbs of transfer and verbs of change of possession with regard to passivization:
(28) Verbs of transfer
a. Ken-ga Eri-ni hanataba-o okut-ta/todoke-ta.
Ken-NOM Eri-to bouquet-ACC send-PAST/deliver-PAST
‘Ken sent/delivered a bouquet to Eri.’
b. *Eri-ga Ken-ni hanataba-o okur-are-ta/todoke-rare-ta.
Eri-NOM Ken-by bouquet-ACC send-PASS-PAST/deliver-PASS-PAST
‘Eri was sent/delivered a bouquet by Ken.’
(29) Verbs of change of possession5
a. Titioya-ga kodomo-ni okozukai-o atae-ta/watas-ita.
father-NOM child-DAT allowance-ACC give-PAST/give-PAST
‘Father gave an allowance to the child.’
b. Kodomo-ga titioya-ni okozukai-o atae-rare-ta/watas-are-ta.
child-NOM father-by allowance-ACC give-PASS-PAST/give-PASS-PAST
‘The child was given an allowance by father.’
The ungrammaticality of (28b) suggests that this -ni is a postposition rather than a case-marker. Note that (28b) is unacceptable in the direct passive, but is actually fine in the indirect passive with an adversative reading on the nominative phrase Eri.
The following contrast in (30) with respect to the -e/-ni alternation also leads to the same conclusion:
4 The movement of the dative possessor in (27) is from a deeply embedded position:, i.e., from out of DP contained in PP. I assume that extraction of the dative possessor out of DP parallels that in Hungarian (Szabolsci 1983), and extraction from within PP is permitted as has been proposed in Van Riemsdijk (1978). However, questions remain as to how such movement is motivated and what syntactic operations are executed.
5 Miyagawa and Tsujioka (2004) report that the verbs in (i) allow double-goal ditransitive constructions.
(i) okuru ʻsend,ʼ todokeru ʻdeliver,ʼ kaesu ʻreturn,ʼ kakeru ʻring,ʼ ataeru, ʻgive,ʼ dasu ʻsendʼ and azukeru ʻentrustʼ
The verb ataeru ʻgiveʼ in (i) is a verb of change of possession, as in (29), which runs counter to my argument. However, I find it difficult to construct with this verb grammatical EPCs of the type discussed in this paper.
(30) Verbs of transfer
a. Ken-ga Eri-ni/-e hanataba-o okut-ta/todoke-ta.
Ken-NOM Eri-to bouquet-ACC send-PAST/deliver-PAST
‘Ken sent/delivered a bouquet to Eri.’
Verbs of change of possession
b. Titioya-ga kodomo-ni/*-e okozukai-o atae-ta/watas-ita.
father-NOM child-DAT allowance-ACC give-PAST/give-PAST
‘Father gave an allowance to the child.’
The -ni phrase in the verbs of transfer sentences can alternate with the -e postpositional phrase in (30a), but the -ni phrase in the verbs of change of possession sentences cannot. These results again show that the -ni phrase for verbs of transfer is a postpositional one. On the basis of the above facts, verbs of transfer are considered to be subcategorized for a postpositional -ni phrase, and the dative possessor in double-goal ditransitive constructions is an extra clausal element. I assume with Miyagawa and Tsujioka (2004) that there is an applicative head below v, which assigns a possessor θ-role to the dative possessor.6 Recall that under my analysis, the possessor role is assigned at the base within the possessum phrase, which renders a possessive relation between the possessor and the possessee. The applicative head in question assigns a θ-role that allows the dative possessor to undergo inter- pretation that Theme is to the possession of the possessor.
3. Conclusion
This paper has shown that the dative possessor in double-goal ditransitive constructions à la Miyagawa and Tsu- jioka (2004) undergoes movement out of the locative possessum phrase. These EPCs do not require inalienabili- ty, just as EPCs in Nez Perce (Deal 2013). While there is empirical support for the movement analysis, there also exists counterevidence against such analysis. The availability of a genitive pronoun in the possessum phrase parallels that in finite control clauses, which supports the base-generation analysis. These contradictory facts can be reconciled by adopting Hornstein’s (1999) movement analysis of control. The dative possessor is assigned a θ-role by the possessee in the base position, which renders a possessor-possessee reading. It is assigned another θ-role by an applicative head below v, which allows an interpretation that the possessor is in the possession of Theme. Given the parallelism between finite control clauses and double-goal ditransitive constructions in Japa- nese, it remains to be seen what the internal syntax of possessum phrases is like. I will leave the question open for future research.
References
Akuzawa, Koyo. 2017. Kontorooru-gensyo-no toogoteki imiteki bunseki: Syubun-doosi-to hobun-keesiki-no taioo-kankee [Syntactic and semantic investigations into control phenomena: Main verbs and complements]. Ph.D. dissertation, Uni- versity of Tsukuba.
Aoyagi, Hiroshi. 2015. On the notion of subjects and double complement unaccusatives in Japanese. Proceedings of the
6 This paper does not discuss the base order of three clausal elements in double-goal ditransitive constructions. See Miyagawa and Tsujioka (2004), Kishimoto (2008) and Ayano (2016) for discussions on this matter.
Seoul International Conference on Generative Grammar 17: 37-52.
Ayano, Seiki. 2016. How low is Low Goal in Japanese. Proceedings of the Seoul International Conference on Generative Grammar 18: 34-48.
Ayano, Seiki. 2017. Dative phrase raising in double-goal ditransitive constructions in Japanese Proceedings of the Seoul International Conference on Generative Grammar 19: 71-84.
Chomsky, Noam. 1995. The minimalist program. Cambridge, MA: MIT Press.
Deal, Amy Rose. 2013. Possessor raising. Linguistic Inquiry 44: 391-432.
Doron, Edit and Caroline Heykock. 1999. Filling and licensing multiple specifiers. Specifiers: Minimalist approaches, ed.
by David Adger et al., 69-89. Oxford: Oxford University Press.
Doron, Edit and Caroline Heykock. 2009. In support of broad subjects in Hebrew. Lingua 120: 1764-1776.
Funakoshi, Kenshi. 2017. Backward control from possessors. Syntax 20: 170-213.
Hasegawa, Nobuko. 2009. Tyokusetu-zyudoobun to syoyuu-zyudoobun: Little v tosite-no -rare to sono sosee [Direct pas- sive and possessive passive: -rare as little v and its features]. Goi-no imi to bunpoo [Lexical meaning and grammar], ed.
by Yoko Yumoto and Hideki Kishimoto, 433-454. Tokyo: Kuroshio.
Heycock, Caroline. 1993. Syntactic predication in Japanese. Journal of East Asian Linguistics 2: 167-211.
Hornstein, Nobert. 1999. Movement and control. Linguistic Inquiry 30: 69-96.
Kikuchi, Akira. 1994. Extraction from NP in Japanese. Current topic in English and Japanese, ed. by Masaru Nakamura, 79-104. Tokyo: Hituzi Syobo.
Kishimoto, Hideki. 2001. The role of lexical meanings in argument encoding: Double object verbs in Japanese. Gengo Kenkyu 120: 35-65.
Kishimoto, Hideki. 2008. Ditransitive idioms and argument structure, Journal of East Asian Linguistics 17: 141-179.
Kishimoto, Hideki. 2009. Subject raising in Japanese. MIT Working Papers in Linguistics 58: 225-240.
Kishimoto, Hideki. 2010. Subjects and constituent structure in Japanese. Linguistics 48: 629-670.
Kishimoto, Hideki. 2013. Covert possessor raising in Japanese. Natural Language and Linguistic Theory 31: 161-205.
Landau, Idan. 1999. Possessor raising and the structure of VP. Lingua 107: 1-37.
Landau, Idan. 2004. The scale of finiteness and the calculus of control. Natural Language & Linguistic Theory 22: 811- 877.
Lee, Kum Young. 2009. Finite control in Korean. Ph.D. dissertation, University of Iowa.
Miyagawa, Shigeru and Takae Tsujioka. 2004. Argument structure and ditransitive verbs in Japanese. Journal of East Asian Linguistics 13: 1-38.
Ogawa, Yoshiki. 2001. The stage/individual level distinction and (in)alienable possession. Language 77: 1-25.
Riemsdijk, Henk van. 1978. A case study in syntactic markedness: The binding nature of prepositional phrases. Dordrecht:
Foris.
Szabolsci, Anna. 1983. The possessor that ran away from home. The Linguistic Review 3: 89-102.
Takezawa, Koichi. 2016. Nihongo moodaru-zyutugo-koobun-no toogo-koozoo-to ziseezi-no toogoteki-yakuwari [Modal predicate constructions and syntactic function of tense in Japanese]. Niti-ee-taisyoo bunpoo-to goi-e-no toogooteki- apurooti: Seesee-bunpoo ninti-gengogaku-to nihongogaku [Japanese-English A comprehensive approach to grammar and lexicon: Generative grammar, cognitive linguistics and Japanese linguistics], ed. by Koji Fujita and Yoshiki Nishimura, 55-76. Tokyo: Kaitakusha.
Tomioka, Satoshi and Chang-Yong Sim. 2007. The event semantic root of inalienable possession. Unpublished Ms, Uni- versity of Delaware.
Uchibori, Asako. 2000. The syntax of subjunctive complements: Evidence from Japanese. Ph.D. dissertation, University of Connecticut.
Ura, Hiroyuki. 1996. Multiple feature-checking: A theory of grammatical feature splitting. Ph.D. dissertation, MIT.
Watanabe, Akira. 1996. Nominative-genitive conversion in Japanese: A crosslinguistic perspective. Journal of East Asian Linguistics 5: 373-410.
Yoon, James. 1990. Theta theory and the grammar of inalienable possession constructions. NELS 20: 502-516.
Yoon, James. 2015. Double nominative and double accusative constructions. Handbook of Korean Linguistics, ed. by Lucien Brown and Jaehoon Yeon, 79-97. New York: John Wiley & Sons.
日本語における新種の所有者句外置構文に関する考察
綾 野 誠 紀
要 旨
諸言語において観察される所有者句外置構文について、基底生成分析と移動分析という相対する分析 が提案されてきた。所有者句外置構文の移動分析への反論として、affectedness conditionが適用される ことが挙げられる。すなわち、所有句と被所有名詞の間に典型的に見られる分離不可能性が所有者句 外置構文の成立要件の一部であることが、移動分析への問題として指摘されている。本稿では、Mi-
yagawa and Tsujioka (2004)が提示した2種の目的句を含む二重目的語構文と類似した構文が、所有者
句外置構文であること、また、その構文の成立に分離不可能性が関与しないことを示す。さらに、所 有者句外置構文が所有者句の移動により派生されている証拠を提示する。同時に移動分析への反例が あることも指摘し、これらの矛盾する事実はHornstein (1999)において提案されているコントロール 構文の移動分析により説明できると主張する。最後に、この所有者句外置構文の統語構造についても 考察する。
