2397-1835 Glossa: a journal of general linguistics 2397-1835 Open Library of Humanities 10.16995/glossa.8850 Research article Phasal transfer and anti-locality: Movement of objects in Japanese Takahashi Masahiko mstakahashi@human.kj.yamagata-u.ac.jp 1 Faculty of Humanities and Social Sciences, Yamagata University, JP 13 11 2023 2023 8 1 1 39 Copyright: © 2023 The Author(s) 2023 This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International License (CC-BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. See http://creativecommons.org/licenses/by/4.0/.

In this paper, based on an analysis of the scope properties of objects in Japanese, I argue that (i) an entire phase is transferred when a higher phase is completed, and (ii) the movement of objects and adjuncts is constrained by anti-locality. I propose that in the potential construction, the movement of accusative objects is more constrained than the movement of nominative objects owing to the interaction between phasal transfer and anti-locality, which leads the accusative and nominative objects to exhibit distinct scope behaviors. This analysis leads us to conclude that negation in Japanese is an adjunct. Furthermore, the analysis correctly predicts the movement of objects in simple transitive sentences and the non-movement of nominative phrases in certain cases.

 1 Introduction

This paper investigates the interaction between phases/transfer domains and the anti-locality of movement. It has been generally assumed that transfer, whose domain is defined by phases (Chomsky 2000), sends relevant semantic and phonological features to the conceptual–intentional and sensory–motor interfaces, respectively, and is applied multiple times during syntactic derivations. Once transferred, syntactic objects are assumed to be inaccessible for further syntactic computation, which reduces the computational load. Since Chomsky (2000), the “standard” hypothesis has been that v*Ps and CPs constitute phases, and what is transferred are phasal complements (VPs and TPs, respectively). However, there is still debate about what should count as phases/transfer domains (Uriagereka 1999; Chomsky 2000; 2008; 2015; Bobaljik & Wurmbrand 2005; Boeckx & Grohmann 2007; Takahashi 2010; 2011; Bošković 2005; 2007; 2014a; 2015; 2016a; b; Saito 2017a; b; 2020; Holmberg et al. 2019; Sheehan & Cyrino to appear, among others). Another related issue is (anti-)locality. Although it is generally accepted that movement cannot be too long, many authors have also argued based on various phenomena that movement cannot be too short (Bošković 1994; 1997; 2005; 2014a; 2016a; b; Ishii 1999; Abels 2003; Grohmann 2003; Funakoshi 2015; Erlewine 2020; Branan 2023, among others). A ban on movement that is too local is called anti-locality (Grohmann 2003); however, its precise definition is still being debated.

This paper aims to advance our understanding of the aforementioned issues by providing a novel analysis of the scope properties of objects in Japanese. Transitive objects in Japanese can receive accusative Case or nominative Case when a transitive predicate is accompanied by a potential suffix, which is [+ stative] (Kuno 1973; Saito 1982; Takezawa 1987):

(1)

Kodomo-tati-ga

child-pl-nom

kanzirensyuu-o/*ga

kanji.practice-acc/nom

tuzuke-ru.

continue-prs

‘Children continue kanji practice.’

(2)

Kodomo-tati-ga

child-pl-nom

kanzirensyuu-o/ga

kanji.practice-acc/nom

tuzuke-rare-ru.

continue-can-prs

‘Children can continue kanji practice.’                                                                                                                                (Takahashi 2021a: 154)

The transitive object kanzirensyuu ‘kanji practice’ in (1) can only receive accusative Case from the transitive verb tuzuke ‘continue’, which is [– stative]. In (2), tuzuke ‘continue’ has a potential suffix -rare ‘can’, which is [+ stative]. The transitive object kanzirensyuu ‘kanji practice’ in (2) can receive either nominative Case or accusative Case.1 Significantly, accusative and nominative objects show distinct scope behaviors (Sano 1985; Tada 1992; 1993; Koizumi 1994; 1998; Saito & Hoshi 1998; Ura 1999; Wurmbrand 2001; Takano 2003; Bobaljik & Wurmbrand 2005; 2007; Nomura 2005; Saito 2010a; 2012; Takahashi 2010; 2011; 2021a; Funakoshi & Takahashi 2014; Shimamura & Wurmbrand 2014; Kasai 2018; Ochi & Isono 2021; Moritake 2022, among others).

The scope properties of nominative objects have received considerable attention in the literature (see Section 2 for an overview). In this paper, I provide further evidence for a previous proposal that a nominative object moves into the TP domain when it takes scope over the potential suffix (Koizumi 1994; 1998; Nomura 2005; Ochi & Isono 2021). Furthermore, given that transitive objects can usually take scope over negation (Kato 1985; Han et al. 2004; Kataoka 2006; Shibata 2015), I argue that what requires more articulated explanation is the obligatory narrow scope behavior of the accusative object in the potential construction, which has received relatively little attention in the literature (cf. Shibata 2015). Specifically, the accusative object in the potential construction fails to move into the TP domain via scrambling owing to the interaction between phasal transfer (Bošković 2016a; Saito 2017a; b; 2020) and anti-locality (Bošković 2005; 2014a; 2016b).

This paper is organized as follows. In Section 2, I present some core observations that the paper addresses and show that previous analyses cannot fully explain them. In Section 3, I provide an analysis of the observations, which crucially relies on the interaction of phasal transfer, anti-locality, and scrambling. In Section 4, I discuss the consequences of the proposed analysis. First, I discuss the status of negation in Japanese and suggest that it is an adjunct, which leads us to an analysis of the scope properties of objects in simple transitive sentences. Second, I discuss cases where nominative phrases apparently undergo movement in violation of the constraints introduced in the previous sections. I show that these nominative phrases do not in fact undergo movement, which lends further credence to the proposed analysis.

 2 Core observations and previous analyses

In this section, I introduce three core observations that are discussed in the following sections, and point out that previous analyses fail to provide a unified account of them. In particular, I discuss (i) accusative/nominative objects that co-occur with nominative subjects, (ii) adjuncts that follow accusative/nominative objects, and (iii) nominative objects that co-occur with instrumental subjects.

First, nominative objects, but not accusative objects, can take scope over the potential suffix when preceded by nominative subjects (Sano 1985; Tada 1992; 1993; Koizumi 1994; 1998; Saito & Hoshi 1998; Ura 1999; Wurmbrand 2001; Takano 2003; Bobaljik & Wurmbrand 2005; 2007; Nomura 2005; Saito 2010a; 2012; Takahashi 2010; 2011; 2021; Funakoshi & Takahashi 2014; Shimamura & Wurmbrand 2014; Kasai 2018; Ochi & Isono 2021; Moritake 2022, among others):2

(3)

Observation 1: scope of accusative/nominative objects

Scenario for the can > only reading: the children are able to concentrate on their kanji practice without engaging in other activities such as listening to music or watching TV.

Scenario for the only > can reading: among several activities such as listening to music, watching TV, and kanji practice, the children cannot continue anything except for kanji practice.

 

a.

Kodomo-tati-ga

child-pl-nom

kanzirensyuu-dake-o

kanji.practice-only-acc

tuzuke-rare-ru.

continue-can-prs

‘Children can continue only kanji practice.’

‘Children can continue kanji practice without continuing anything else.’ (can > only)

‘It is only kanji practice that children can continue.’ (?*only > can)

 

b.

Kodomo-tati-ga

child-pl-nom

kanzirensyuu-dake-ga

kanji.practice-only-nom

tuzuke-rare-ru.

continue-can-prs

‘Children can continue only kanji practice.’

‘Children can continue kanji practice without continuing anything else.’ (can > only)

‘It is only kanji practice that children can continue.’ (only > can)                                                                   (Takahashi 2021a: 154)

Whereas the accusative object in (3a) must take scope under the potential suffix, the nominative object in (3b) can take scope either under or over the potential suffix (see Nomura 2005; Takahashi 2010; 2011; 2021a; Funakoshi & Takahashi 2014; Shimamura & Wurmbrand 2014; Kasai 2018; Ochi & Isono 2021 for the availability of the narrow scope interpretation of the nominative object; see below for further evidence).

Second, the Case of the object affects the scope of dake ‘only’ contained in an adjunct (Saito & Hoshi 1998; Bobaljik & Wurmbrand 2007; Takahashi 2010; 2011. cf. Takano 2003).

(4)

Observation 2: scope of adjuncts following accusative/nominative objects

Scenario for the can > only reading: the children have the ability to continue kanji practice with only a pen; they do not need pencils, notes, sheets of papers, or anything else that might be used for kanji practice.

Scenario for the only > can reading: among writing materials that could be used for kanji practice, the children can only use a pen, and they cannot use any other tools for kanji practice.

 

a.

Kodomo-tati-ga

child-pl-nom

kanzirensyuu-o

kanji.practice-acc

pen-dake-de

pen-only-with

tuzuke-rare-ru.

continue-can-prs

‘Children can continue kanji practice with only a pen.’

‘Children can continue kanji practice with a pen and nothing else.’ (can > only)

‘It is only a pen that children can continue kanji practice with.’ (?*only > can)

 

b.

Kodomo-tati-ga

child-pl-nom

kanzirensyuu-ga

kanji.practice-nom

pen-dake-de

pen-only-with

tuzuke-rare-ru.

continue-can-prs

‘Children can continue kanji practice with only a pen.’

‘Children can continue kanji practice with a pen and nothing else.’ (?can > only)

‘It is only a pen that children can continue kanji practice with.’ (only > can)

Dake ‘only’ in (4a) and (4b) is contained in the adjunct. It takes scope under the potential suffix when the object receives accusative Case (4a). It can also take scope over the potential suffix when the object receives nominative Case (4b).3

Third, the scope of the objects is affected by subject marking. I show below that nominative objects do not take scope over potential suffixes in the presence of instrumental subjects (Ebina 2020; Takahashi 2021a). In Japanese, subjects can appear with the instrumental marker -de, which is a postposition used to mark instruments (e.g., naifu-de ‘with a knife’; see Takubo 1984; Inoue 1998; Kishimoto 2005; 2010; 2017). Following Kishimoto (2005; 2010), I use the term instrumental subjects to refer to subjects marked with -de (see Kishimoto 2010; 2017 for evidence of the subjecthood of instrumental subjects). Significantly, nominative objects must take scope under the potential suffix when subjects appear with -de ‘with’ (Ebina 2020; Takahashi 2021a):4

(5)

Observation 3: scope of nominative objects following instrumental subjects

 

a.

Kodomo-tati-ga

child-pl-nom

kanzirensyuu-dake-ga

kanji.practice-only-nom

tuzuke-rare-ru.

continue-can-prs

‘Children can continue only kanji practice.’

‘Children can continue kanji practice without continuing anything else.’ (can > only)

‘It is only kanji practice that children can continue.’ (only > can) (= (3b))

 

b.

Kodomo-tati-de

child-pl-with

kanzirensyuu-dake-ga

kanji.practice-only-nom

tuzuke-rare-ru.

continue-can-prs

‘Children can continue only kanji practice.’

‘Children can continue kanji practice without continuing anything else.’ (can > only)

‘It is only kanji practice that children can continue.’ (?*only > can)                                                                     (Takahashi 2021a: 156)

In (5a), where the subject receives nominative Case, the following nominative object can take scope over the potential suffix. However, in (5b), where the subject is accompanied by the instrumental marker -de ‘with’, the nominative object must take scope under the potential suffix.5

Let us now consider how previous analyses fare with the above observations. Here, I focus on two approaches: (i) the Case-movement analysis and (ii) the Quantifier Raising (QR) analysis.6 A well-known analysis of the scope puzzle is what I call the Case-movement analysis (Tada 1992; 1993; Koizumi 1994; 1998; Nomura 2005; Ochi & Isono 2021, among others). This analysis suggests that objects may appear in distinct positions depending on their Case, which immediately captures the first observation (3). In this analysis, the accusative object in (3a) remains within the vP complement selected by the potential suffix, which is responsible for the narrow scope behavior of the accusative object:

(6)

Case-movement analysis: accusative object (3a)

In (6a), the vP phase is constructed. The embedded object receives accusative Case from the v-head (Chomsky 2000; 2008; 2015).7 In (6b), the VP complement is transferred.8 In (6c), the potential suffix and the subject are introduced into the derivation, constructing the vcanP. In (6d), the nominative subject moves to Spec, TP. As the accusative object remains within the vP, it takes scope under the potential suffix. However, the nominative object in (3b) can move into the TP domain, which is responsible for the wide scope behavior of the nominative object. For the sake of exposition, I assume here that the potential suffix directly selects the VP complement (Wurmbrand 2001; Bobaljik & Wurmbrand 2005; 2007; Nomura 2005):

(7)

Case-movement analysis: nominative object (3b)

 

a.

Constructing the TP and nominative Case assignment:

[TP                                             [vcanP    SUBJNOM    [VP    OBJNOM    V]    vcan]    T]

 

b.

Movement of the nominative phrases:

[TP    SUBJiNOM    OBJjNOM    [vcanP       ti                [VP       tj               V]    vcan]    T]

In (7a), the subject and the object receive nominative Case from T (Hiraiwa 2001; Nomura 2005; Takahashi 2010; 2011; 2021a; Ochi & Isono 2021, among others).9 In (7b), the subject and the object move into the TP domain (Koizumi 1994; 1998; Nomura 2005; Ochi & Isono 2021, among others). Therefore, the nominative object can take scope over the potential suffix in the TP domain.10

The Case-movement analysis can also account for the third observation (5) concerning subject marking if we assume that (i) nominative objects can remain in their base-generated positions (Nomura 2005; Ochi & Isono 2021) and (ii) whereas nominative subjects move to Spec, TP, instrumental subjects remain in their base-generated positions (i.e., they do not undergo movement to Spec, TP; Ebina 2020; Takahashi 2021a; cf. Fukui 1986; Kuroda 1988; Kishimoto 2010; 2017; see Section 3.2 for a concrete analysis of instrumental subject constructions).

(8)

Case-movement analysis:

 

a.

Nominative object and nominative subject (5a):

[TP    SUBJiNOM    OBJjNOM    [          ti              [VP      tj              V]    vcan]    T]

 

b.

Nominative object and instrumental subject (5b):

[TP                                             [       SUBJINS    [VP    OBJNOM    V]   vcan]    T]

In (8a), the nominative subject moves to Spec, TP, and the following nominative object can also move into the TP domain, taking scope over the potential suffix. In (8b), the instrumental subject remains in its base-generated position, and the following nominative object must remain in the VP, taking scope under the potential suffix.11

However, under the Case-movement analysis, the second observation regarding adjuncts (4) remains unexplained because dake ‘only’ is contained in the adjuncts, which do not undergo Case-movement (Saito & Hoshi 1998; Bobaljik & Wurmbrand 2007; Takahashi 2010; 2011. cf. Takano 2003):

(9)

Case-movement analysis:

 

a.

Accusative object (4a):

[TP    SUBJiNOM    [vcanP  ti      [vP      PRO    [VP    OBJACC    Adjunct(only)    V]    v] vcan] T]

 

b.

Nominative object (4b):

[TP    SUBJiNOM    OBJjNOM    [vcanP    ti       [VP      tj             Adjunct(only)     V]   vcan] T]

As the Case-movement analysis attributes the scope contrast to the positions of the Case-marked objects, it remains unclear why the scope of dake ‘only’ contained in the adjunct is affected by the (non)-movement of the objects. In particular, while the narrow scope behavior of dake ‘only’ in (9a) may be accounted for by the adjunct’s position below the potential suffix,12 it remains unclear why dake ‘only’ can take scope over the potential suffix in (9b).

The other analysis discussed here is what I call the QR-analysis (Bobaljik & Wurmbrand 2007; Takahashi 2010; 2011). I examine Takahashi’s (2010; 2011) analysis, which suggests that dake ‘only’ alone undergoes Quantifier Raising (Bobaljik & Wurmbrand 2007), which is bound to domains of Case-valuation.13 The QR analysis can accommodate the first observation (3) because the accusative object and the nominative object receive the Case from distinct probes (note that Takahashi (2010; 2011) assumes that the potential suffix selects vP regardless of the Case of the embedded object):

(10)

QR analysis:

In (10a), the object receives accusative Case from the embedded v. Takahashi (2010; 2011) assumes that dake ‘only’ has to adjoin to the node of type t and that further movement of dake ‘only’ is impossible after the movement for type-resolution. Dake ‘only’ contained in the accusative object thus moves to the embedded vP and takes scope there, which captures the obligatory narrow scope interpretation of dake ‘only’. In (10b), the embedded object (and the subject) receive nominative Case from T. As the embedded v does not assign accusative Case to the object, dake ‘only’ contained in the object can move beyond the embedded vP. Dake ‘only’ takes scope over the potential suffix by adjoining to the TP projection and takes scope under the potential suffix by adjoining to the vP complement.

This QR analysis naturally captures the second observation concerning adjuncts (4), which is problematic for the Case-movement analysis, because what undergoes movement is not the Case-marked NPs but the scope-bearing element dake ‘only’:

(11)

QR analysis:

In (11a), the object receives accusative Case from the embedded v, and dake ‘only’ contained in the adjunct thus moves to the embedded vP and takes scope there. Accordingly, dake ‘only’ takes scope under the potential suffix. In (11b), the embedded object receives nominative Case from T. As the embedded v does not assign accusative Case to the object, dake ‘only’ in the adjunct can move beyond the embedded vP and can take scope over or under the potential suffix when it adjoins to the TP or the vP, respectively.

However, the third observation concerning instrumental subjects (5) remains problematic because it is at least unclear why in-situ subjects block the QR of dake ‘only’ (I assume that the instrumental subject is in Spec, vP; see Section 3.2):

(12)

QR analysis:

In both (12a) and (12b), the embedded object receives nominative Case from T. Hence, the scope behavior of dake ‘only’ contained in the object should be the same regardless of the Case of the subject (unless we stipulate that the instrumental subject somehow blocks movement of dake ‘only’).

Let me summarize the results obtained thus far:

(13)

 

Observation 1 (object) (3)

Observation 2 (adjunct) (4)

Observation 3 (instrumental subject) (5)

Case-movement analysis

            ✔  (6)/(7)

            *    (9)

            ✔  (8)

QR analysis

      ✔ (10)

      ✔ (11)

      *   (12)

As shown above, the two analyses reviewed have different empirical coverage. Although the Case-movement analysis can account for Observation 3 concerning instrumental subjects, it cannot explain Observation 2 concerning adjuncts. In contrast, although the QR analysis can account for Observation 2, it fails to explain Observation 3. Hence, we need a “hybrid” analysis that combines the insights of both approaches. Specifically, we need to ensure that (i) the scope properties of objects reflect their positions (as in the Case-movement analysis for Observation 3) and (ii) a scope-bearing element contained in an adjunct can be located in a high position in certain cases (as in the QR analysis for Observation 2). Furthermore, there is one important observation that requires consideration: it has been shown that objects in Japanese generally take a wide scope over negation (Kato 1985; Han et al. 2004; Kataoka 2006; Shibata 2015). Therefore, what seems to require a more articulated explanation is the obligatory narrow scope behavior of accusative objects in the potential construction, rather than the wide scope property of nominative objects.

 3. An analysis: Phasal transfer, anti-locality, and scrambling as adjunction

In this section, I provide a new analysis that accounts for the three observations discussed in the previous section. First, I introduce some theoretical ingredients of the new analysis: phasal transfer, anti-locality, the condition on multiple phase edges, and scrambling as adjunction. Then, I provide a concrete analysis of the three observations.

3.1 Theoretical ingredients 3.1.1 Phasal transfer

Since Chomsky (2000), there have been various approaches to phases and transfer domains (Uriagereka 1999; Chomsky 2000; 2008; 2015; Bobaljik & Wurmbrand 2005; Boeckx & Grohmann 2007; Takahashi 2010; 2011; Bošković 2005; 2007; 2014a; 2015; 2016a; b; Saito 2017a; b; 2020; Holmberg et al. 2019; Sheehan & Cyrino to appear, among others). Among them, Saito (2017a; b; 2020) proposes the following hypothesis (see also Chomsky 2000 and Bošković 2016a):

(14)

A phase is transferred upon the completion of the next phase up.                                                                                      (Saito 2017a: 64)

This hypothesis captures the well-known difference between English and Japanese concerning the locality of reflexive binding (Yang 1983), and the analysis assumes that Japanese lacks phi-feature agreement:

(15)

a.

*Mary insisted that herself saw it.

 

b.

  Hanako-wa

  Hanako-top

[CP

 

[TP

 

zibunzisin-ga

self-self-nom

sore-o

it-acc

mi-ta]

see-pst

to]

comp

syutyoosi-ta.

insist-pst

  ‘Hanako insisted that she (= Hanako) saw it.’                                                                                                                (Saito 2017a: 61)

As shown in (15a), reflexive binding in English is impossible across a finite clause boundary; the reflexive herself, which is the embedded subject, cannot refer to the matrix subject Mary. In contrast, as shown in (15b), reflexive binding in Japanese is possible across a finite clause boundary; the reflexive zibunzisin ‘self-self’ refers to the matrix subject Hanako. Saito (2017a; b; 2020) argues that the hypothesis in (14) provides a principled account of this contrast. Assuming that (i) a reflexive must be c-commanded by its antecedent at the end of each phase before transfer (Quicoli 2008) and (ii) T and V inherit phasehood from C and v* as well as uninterpretable phi-features in languages with phi-feature agreement (cf. Chomsky 2008; 2015), Saito (2017a; b; 2020) analyzes (15a) and (15b) as follows:

(16)

In (16a), the embedded T inherits uninterpretable phi-features from the embedded C, which makes the embedded TP a phase. Once the embedded CP is completed, the embedded TP is transferred. As the reflexive herself is transferred before the matrix subject Mary c-commands it, herself cannot refer to Mary. In (16b), the embedded T does not inherit uninterpretable phi-features from the embedded C (owing to the lack of uninterpretable phi-features). Therefore, the embedded TP is not a phase. When the embedded CP is completed, the embedded v*P, but not the embedded TP, is transferred. The CP complement is not transferred until the matrix vP is completed. As the matrix subject c-commands the embedded reflexive before the reflexive is transferred, the reflexive zibunzisin ‘self-self’ can refer to the matrix subject Hanako. The analysis presented below extends Saito’s (2017a; b; 2020) hypothesis to the potential construction.14

 3.1.2 Anti-locality

The second theoretical assumption employed in this paper concerns anti-locality. Many authors have claimed that movement cannot be too local (Bošković 1994; 1997; 2005; 2014a; 2016a; b; Ishii 1999; Abels 2003; Grohmann 2003; Funakoshi 2015; Erlewine 2020; Branan 2023). In this paper, I adopt the version of anti-locality proposed by Bošković (2005: 16):

(17)

Each chain link must be at least of length 1, where a chain link from A to B is of length n if there are n XPs that dominate B but not A.

The condition in (17) requires movement to take place across a full phrase (i.e., a segment of a phrase does not count as a full phrase). Bošković (2005) claims that the anti-locality condition in (17) and the phase impenetrability condition (PIC), which dictates that only a phase head and its edges are accessible for movement targeting positions outside the phase (Chomsky 2000), explain the impossibility of left-branch extraction in English:

(18)

*Beautifuli he saw [ti houses].                                                                                (Bošković 2005: 2)

Bošković (2005) claims that the impossibility is explained by an interaction of anti-locality and the PIC:

(19)

The adjective beautiful in (19) is adjoined to the NP, which is dominated by the DP. Assuming that DP is a phase in English, the PIC requires that the adjective be inaccessible for movement to a position outside the DP complement unless the adjective moves to Spec, DP. However, the movement of the adjective to Spec, DP is prohibited by anti-locality because it does not cross a full phrase boundary. Thus, the impossibility of left-branch extraction in English is explained by the interaction between the PIC and the anti-locality condition on movement.

 3.1.3 Condition on multiple phase edges

Another important component of the analysis proposed below concerns the contextuality of phase edges. Whereas there have been proposals in the literature that the phasehood of a particular phrase is determined contextually (Bobaljik & Wurmbrand 2005; Bošković 2005; 2014a; Takahashi 2010; 2011), Bošković (2014b; 2016a) proposes that phase edges must also be determined contextually. Specifically, essentially following Wurmbrand (2013) and Bošković (2014b; 2016a), I assume the following condition on phase edges, which I dub the condition on multiple phase edges:

(20)

Condition on multiple phase edges

When there is more than one element at a phase edge, only the highest element is accessible to operations outside of the phase.

For instance, when a phase XP contains multiple specifiers/adjuncts, only the highest element is accessible for movement and feature-valuation:

(21)

[XP    ✔AP    *BP    [X’    CP    X]]

The AP and BP in (21) are located at the edge of the XP phase. As the AP is located above the BP, the AP, but not the BP, is accessible for operations outside the XP phase. Furthermore, I assume with Wurmbrand (2013) and Bošković (2014b) that the multiple edge condition is evaluated at PF. When an element in the lower specifier/adjunct at the phase edge moves across a higher element at the edge, the latter receives a *, which causes a crash at PF. However, a PF crash can be avoided if the offending higher specifier/adjunct is turned into a lower copy of the movement, which is deleted at PF.

(22)

[YP    AP    BP    [XP    *AP    BP    [X’    CP    X]]]

In (22), the BP on the lower edge of the XP phase moves across the AP on the higher edge of the XP phase, resulting in the *-marking of the AP. However, as the AP also moves, the lower copy of the AP, which contains the * responsible for the PF crash, is now deleted. Hence, there is no PF crash.

 3.1.4 Scrambling as an adjunction to an outer edge

Finally, following previous analyses, I make several assumptions concerning scrambling. I assume, along with Saito (1985; 1989; 1992), Abe (1993), Tada (1993), and Funakoshi (2015), among others, that scrambling is an adjunction operation. Moreover, following Takano (2010) and Takita (2010), I assume that scrambling is not a tucking-in operation in the sense of Richards (2001). Taking this all together, we see that when scrambling of an object targets a vP that involves an external argument, the object must be located at the outer vP-adjoined position:15

(23)

[TP [vP    OBJiACC    [vP    SUBJ    [VP    ti    V] v]] T]

Given the anti-locality constraint in (17), the object in the vP-adjoined position is too close to the TP, which will be a crucial component of the analysis in the next section.

 3.2 Analysis

I now provide an analysis of the key observations discussed in Section 2. First, I provide an analysis of the potential constructions, after which I provide a complete analysis of the three observations discussed above. A relevant example of the potential construction is repeated below:

(24)

Kodomo-tati-ga

child-pl-nom

kanzirensyuu-o/ga

kanji.practice-acc/nom

tuzuke-rare-ru.

continue-can-prs

‘Children can continue kanji practice.’ (= (2))

In the presence of a potential suffix, the embedded object can receive either accusative or nominative Case. I assume that when a phase head X is adjoined to another head Y (i.e., Y is pair-merged with X), X loses its status as a phase head and ceases to assign Case (Epstein et al. 2016; see also Nomura 2020 and Saito 2020 for relevant discussions). I assume the following structures for (24) (Moritake 2022):

(25)

a.

b.

[TP

[TP

SUBJiNOM

SUBJiNOM

[vcanP

[vcanP

ti

ti

[vP

 

PRO

 

[VP

[VP

OBJACC

OBJNOM

V]

V]

vACC]

 

vcan]

v-vcan]

T]

TNOM]

In (25a), the potential suffix selects a vP complement, where the object receives accusative Case. In (25b), the embedded v and the potential suffix are externally pair-merged; consequently, the embedded v ceases to be a phase head that assigns Case (see Ura 1999; Takahashi 2010; 2011; Moritake 2022). Hence, the object receives nominative Case from T. This analysis reconciles two apparently conflicting insights in earlier studies regarding the potential suffix. That is, whereas Ura (1999) and Takahashi (2010; 2011) assume that vcan selects vP, Wurmbrand (2001), Nomura (2005), and Bobaljik & Wurmbrand (2005; 2007) assume that vcan directly selects VP. According to the analysis presented above, vcan is directly associated with both the VP complement and the v-head. Additionally, I assume that Case assignment is not contingent on phi-feature agreement (Bošković 2007; Saito 2018).

Having spelled out the relevant assumptions concerning the potential construction, we can return to the first observation, repeated below:

(26)

Observation 1: scope of accusative/nominative objects

 

a.

Kodomo-tati-ga

child-pl-nom

kanzirensyuu-dake-o

kanji.practice-only-acc

tuzuke-rare-ru.

continue-can-prs

‘Children can continue only kanji practice.’

‘Children can continue kanji practice without continuing anything else.’ (can > only)

‘It is only kanji practice that children can continue.’ (?*only > can)

 

b.

Kodomo-tati-ga

child-pl-nom

kanzirensyuu-dake-ga

kanji.practice-only-nom

tuzuke-rare-ru.

continue-can-prs

‘Children can continue only kanji practice.’

‘Children can continue practice kanji practice without continuing anything else.’ (can > only)

‘It is only kanji practice that children can continue.’ (only > can) (= (3))

The nominative object, but not the accusative object, can take scope over the potential suffix. Assuming that objects can move into the TP domain via scrambling in other contexts, I propose to exclude scrambling of the accusative object in (26a). I claim that the movement of the accusative object violates anti-locality and assume that both the vP and the vcanP project phases here:16

(27)

Movement of the accusative object:

In (27a), the vP phase is constructed, and the embedded object receives accusative Case. In (27b), the potential suffix and the subject are introduced into the derivation, constructing the vcanP phase. According to the theory of phasal transfer (14), the vP phase is transferred upon the completion of the vcanP phase; thus, the accusative object adjoins to the outer vcanP edge via scrambling (see Section. 3.1.4), as shown in (27c). Given the condition on multiple phase edges in (20), the object in (27c) must move out of the vcanP so that the subject can move to Spec, TP and receive nominative Case from T. In (27d), the subject and the object move into the TP domain. However, the movement of the object is prohibited owing to anti-locality (17); although the subject movement does not violate anti-locality (i.e., the subject crosses the full vcanP), the movement of the object does violate anti-locality (only the higher segment of the vcanP is crossed). Thus, the accusative object cannot move to a position above the potential suffix, and it remains within the vP complement:

(28)

The object stays in the base-position and takes scope under the potential suffix.

The wide scope interpretation of the nominative object in (26b) also follows from this analysis. The movement of the nominative object does not violate anti-locality.

(29)

Movement of the nominative object:

 

a.

Constructing the TP and nominative Case assignment:

[TP                                             [vcanP    SUBJNOM    [VP    OBJNOM    V]    v-vcan] T]

 

b.

Movement of the nominative phrases:

[TP    SUBJiNOM    OBJjNOM    [vcanP        ti               [VP       tj             V]    v-vcan] T]

In (29a), the subject and the object receive nominative Case from T. In (29b), the subject and the object move into the TP domain without violating anti-locality. The subject moves across the vcanP, and the object moves across the vcanP and the VP. The object can thus take scope over the potential suffix.17

Let us now consider how the current analysis accounts for the second observation regarding adjuncts:

(30)

Observation 2: scope of adjuncts following accusative/nominative objects

 

a.

Kodomo-tati-ga

child-pl-nom

kanzirensyuu-o

kanji.practice-acc

pen-dake-de

pen-only-with

tuzuke-rare-ru.

continue-can-prs

‘Children can continue kanji practice with only a pen.’

‘Children can continue kanji practice with a pen and nothing else.’ (can > only)

‘It is only a pen that children can continue kanji practice with.’ (?*only > can)

 

b.

Kodomo-tati-ga

child-pl-nom

kanzirensyuu-ga

kanji.practice-nom

pen-dake-de

pen-only-with

tuzuke-rare-ru.

continue-can-prs

‘Children can continue kanji practice with only a pen.’

‘Children can continue kanji practice with a pen and nothing else.’ (?can > only)

‘It is only a pen that children can continue kanji practice with.’ (only > can) (= (4))

Dake ‘only’ is contained in the adjunct and takes scope under the potential suffix when the object receives accusative Case, as shown in (30a). When the object receives nominative Case, as shown in (30b), dake ‘only’ contained in the adjunct can take scope over the potential suffix. Under the current analysis, dake ‘only’ in (30a) cannot take scope over the potential suffix because the movement of the accusative object and the adjunct violates anti-locality.

(31)

Movement of the accusative object and the VP adjunct:

In (31a), the object receives accusative Case within the vP phase. In (31b), the potential suffix and the subject are introduced into the derivation, constructing the vcanP phase. In (31c), the object and the adjunct move to the vcanP edge, and the vP phase is transferred. In (31d), the movement of the object and the adjunct violates anti-locality. Hence, the adjunct and the object remain in their base-positions, as shown in (32):

(32)

The movement of the subject does not violate anti-locality. As the adjunct stays below the potential suffix, dake ‘only’ contained in the adjunct must take scope under the potential suffix.

Regarding (30b), the dake ‘only’ contained in the adjunct can take scope over the potential suffix because the adjunct can move into the TP domain via scrambling without violating anti-locality:

(33)

Movement of the nominative object and the adjunct:

In (33a), the subject and the object receive nominative Case from T. In (33b), the subject, the object, and the adjunct move into the TP domain (cf. Shibata 2015). There is no anti-locality violation. The subject moves across the vcanP, and the object and the adjunct move across the vcanP and the VP. Thus, dake ‘only’ can take scope over the potential suffix. Note that the movement of the adjunct (and the object) is optional; if the adjunct stays in its base-position, dake ‘only’ contained in the adjunct takes scope under the potential suffix:18,19

(34)

Non-movement of the adjunct:

Finally, let us consider the third observation concerning instrumental subjects. Before examining the relevant examples, I provide an analysis of instrumental subject constructions, as exemplified below:

(35)

Kodomo-tati-de

child-pl-with

kanzirensyuu-o

kanji.practice-acc

tuzuke-ru.

continue-prs

‘Children continue kanji practice.’

The subject kodomo-tati ‘children’ in (35) appears with the instrumental marker -de, and the object kanzirensyuu ‘kanji practice’ receives accusative Case. I propose that (i) instrumental subjects are base-generated in the lower specifier of an (agentive) v and (ii) a pro base-generated in the higher specifier of the agentive v moves to Spec, TP.

(36)

[TP    proi    [vP    ti    [vP    SUBJINS …    v]    ]    T]

There are two thematic subject positions in (36), and only the instrumental subject is overtly realized (I assume, following Saito (2017b; 2017c), that two NPs can receive the same theta-role from a predicate in violation of the theta-criterion (see also Kuroda 1988)). The instrumental subject remains within the vP. The analysis in (36) makes several predictions that are borne out. First, the analysis predicts that the higher thematic subject may be overtly realized, which is confirmed by examples involving “subject doubling”. Interestingly, nominative subjects and instrumental subjects can co-occur (cf. Takubo 1984; Kishimoto 2017; see Saito 2017b; c; Kuroda 1988 for doubling of arguments):

(37)

Kodomo-tati-ga

child-pl-nom

otokonoko-tati-de

boy-pl-with

kanzirensyuu-o

kanji.practice-acc

tuzuke-ta.

continue-pst

‘Lit. Children are such that boys continued kanji practice.’

In (37), kodomo-tati ‘children’ receives nominative Case, and otokonoko-tati ‘boys’ appears with the instrumental marker -de ‘with’. This example is fully acceptable.

Second, both the nominative and instrumental phrases under consideration behave as genuine subjects (cf. Kishimoto 2010; 2017). This is demonstrated by the following example, which involves the subject-oriented reflexive zibun-tati-zisin ‘self-PL-self’:

(38)

Kodomo-tatii-ga

child-pl-nom

otokonoko-tatij-de

boys-pl-with

zibun-tati-zisini/j-no-hon-o

self-pl-self-gen-book-acc

sute-ta.

discard-pst

‘Lit. Childreni are such that boysj discarded theiri/j books.’

In (38), either kodomo-tati ‘children’ or otokonoko-tati ‘boys’ can be the antecedent of the reflexive zibun-tati-zisin ‘self-pl-self’. When kodomo-tati ‘children’ is chosen as the antecedent, (38) means that the boys discarded books owned by the children, including the boys. However, when otokonoko-tati ‘boys’ is chosen as the antecedent, (38) denotes that the boys discarded books owned by the boys. Assuming that Spec, vP counts as the subject position responsible for the binding of subject-oriented reflexives (Saito 2006), the ambiguity of (38) shows that both the nominative phrase kodomo-tati ‘children’ and the instrumental phrase otokonoko-tati ‘boys’ are located in Spec, vP.

Finally, instrumental subjects cannot be selected by stative predicates, which suggests that instrumental subjects must be associated with agentive/volitional predicates (Takubo 1984; Kishimoto 2005). Instrumental subjects are not allowed with wakar- ‘understand’:

(39)

Watasi-tati-ga/ni/?*de

we-pl-nom/dat/with

eego-ga

English-nom

wakar-u.

understand-prs

‘We understand English.’

Wakar- ‘understand’ usually takes a nominative object, which is available when a predicate is [+ stative] (Kuno 1973; Saito 1982). Although the nominative and the dative subject can co-occur with wakar- ‘understand’, the instrumental subject cannot appear with it. The contrast indicates that instrumental subjects are not available with stative predicates.

We are now ready to discuss the derivation of the potential construction with the instrumental subject:

(40)

Observation 3: scope of nominative objects following instrumental subjects

 

a.

Kodomo-tati-ga

child-pl-nom

kanzirensyuu-dake-ga

kanji.practice-only-nom

tuzuke-rare-ru.

continue-can-prs

‘Children can continue only kanji practice.’

‘Children can continue kanji practice without continuing anything else.’ (can > only)

‘It is only kanji practice that children can continue.’ (only > can) (= (5a))

 

b.

Kodomo-tati-de

child-pl-with

kanzirensyuu-dake-ga

kanji.practice-only-nom

tuzuke-rare-ru.

continue-can-prs

‘Children can continue only kanji practice.’

‘Children can continue kanji practice without continuing anything else.’ (can > only)

‘It is only kanji practice that children can continue.’ (?*only > can) (= (5b))

Whereas the nominative object in (40a) follows the nominative subject and can take scope over the potential suffix, the nominative object in (40b) follows the instrumental subject and must take scope under the potential suffix. Recall that the instrumental subject is located in the lower specifier of the agentive v in (39). This suggests that the instrumental subject cannot be in the specifier of the potential suffix, which is a stative predicate (contra Takahashi 2021a). Hence, I assume that the instrumental subject is located in the lower specifier of the embedded agentive v:

(41)

a.

b.

[TP SUBJiNOM

[TP proi  [vcanP

OBJjNOM  [vcanP     ti

ti   [vP PROi  SUBJINS    [VP

     [VP

OBJNOM

tj

V]

V]

tj]

v-vcan]

vj-vcan]

TNOM] (= (40a))

TNOM] (= (40b))

In (41a), the embedded v is externally adjoined to the potential suffix vcan (i.e., the potential suffix is externally pair-merged with the embedded v). Both the nominative subject and the nominative object move into the TP domain without violating anti-locality (see (29b)). In (41b), the instrumental subject is base-generated in the lower specifier of the embedded vP, and the higher specifier is realized as a PRO, which is controlled by the pro selected by the potential suffix. The embedded v is adjoined to the potential suffix via movement (i.e., the potential suffix is pair-merged with the embedded v) and ceases to be a phase head that assigns accusative Case (cf. Bošković 2015; 2016a). Hence, the embedded object receives nominative Case from T. As the object in (41b) follows the instrumental subject, which remains within the embedded vP, it must take scope under the potential suffix.

The proposed analysis is supported by several observations. First, as the instrumental subject is located below the potential suffix, the former must take scope under the latter. This prediction is borne out:

(42)

a.

Kodomo-tati-dake-ga

child-pl-only-nom

kanzirensyuu-ga

kanji.practice-nom

tuzuke-rare-ru.

continue-can-prs

‘Only children can continue kanji practice.’

‘Children can continue kanji practice without any other people around.’ (*can > only)

‘It is only children who can continue kanji practice.’ (only > not)

 

b.

Kodomo-tati-dake-de

child-pl-only-with

kanzirensyuu-ga

kanji.practice-nom

tuzuke-rare-ru.

continue-can-prs

‘Only children can continue kanji practice.’

‘Children can continue kanji practice without any other people around.’ (can > only)

‘It is only children who can continue kanji practice.’ (*only > not)                                                                  (Takahashi 2021a: 165)

Whereas the nominative subject in (42a) must take scope over the potential suffix, the instrumental subject in (42b) must take scope under the potential suffix. I assume that the obligatory wide scope interpretation of the nominative subject in (42a) is attributed to the well-known observation that sentence-initial nominative phrases with an individual predicate must receive an exhaustive-listing interpretation (Kuno 1973). Importantly, the obligatory narrow scope interpretation of the instrumental subject in (42b) follows from the current analysis because the instrumental subject is located below the potential suffix.

Furthermore, the proposed analysis predicts that the accusative object should be able to move to a position above the nominative subject, taking scope over the potential suffix:

(43)

Movement of the accusative object:

In (43a), the object moves to the edge of the vcanP phase, and the vP phase is transferred. In (43b), the nominative subject moves to Spec, TP, and the accusative object moves into the CP domain, which does not violate anti-locality (the object moves across the TP).20 Recall that if the higher specifier/adjunct is turned into a lower copy of the movement, a violation of the condition on multiple phase edges (20) can be avoided (see Section 3.1.3). As the higher vP edge is the trace/lower copy of the accusative object, the movement of the nominative subject does not violate the condition on multiple phase edges.

(44)

Kanzirensyuui-dake-o

kanji.practice-only-acc

kodomo-tati-ga

child-pl-nom

ti

 

tuzuke-rare-ru.

continue-can-prs

‘Children can continue only kanji practice.’

‘Children can continue kanji practice without continuing anything else.’ (can > only)

‘It is only kanji practice that children can continue.’ ((?)only > can)

In (44), the accusative object in sentence-initial position can take scope over the potential suffix, as predicted by the present analysis.21

To summarize, I have provided an analysis of the core observations discussed in the previous section. The accusative object (and the following adjunct) fail to move into the TP domain because of an interaction between phasal transfer and anti-locality, whereas the nominative object (and the following adjunct) can move into the TP domain without violating anti-locality. Thus, the analysis provides new evidence in favor of phasal transfer and anti-locality.22

 4 Consequences

In this section, I explore some consequences of the present analysis. First, I suggest that negation in Japanese is an adjunct and extend the current analysis to simple transitive sentences, where accusative objects take scope over negation. I then discuss cases where nominative phrases apparently undergo illicit movement into the TP domain. I argue that such cases should be analyzed in terms of the base-generation of the nominative phrases, which provides further credence to the current analysis.

4.1 Negation as an adjunct

In this subsection, I discuss the implications of the proposed analysis for the status of negation.23 The accusative object in the potential construction must take scope under negation, whereas the nominative object can take scope over negation (Koizumi 1994; 1998; Nomura 2005):

(45)

a.

Kodomo-tati-ga

child-pl-nom

kanzirensyuu-dake-o

kanji.practice-only-acc

tuzuke-rare-na-i.

continue-can-neg-prs

‘Children cannot continue only kanji practice.’

‘Children cannot continue kanji practice without continuing anything else.’ (not > only)

‘It is only kanji practice that children cannot continue.’ (?*only > not)

 

b.

Kodomo-tati-ga

child-pl-nom

kanzirensyuu-dake-ga

kanji.practice-only-nom

tuzuke-rare-na-i.

continue-can-neg-prs

‘Children cannot continue only kanji practice.’

‘Children cannot continue kanji practice without continuing anything else.’ (not > only)

‘It is only kanji practice that children cannot continue.’ (only > not)

Although the accusative object in (45a) must take scope under negation, the nominative object in (45b) can take scope over negation. The problem here is that the accusative object may take scope over negation (and the potential suffix) when there is a NegP projection above the vcanP:

(46)

In (46a), the object moves to the edge of the vcanP phase, and the vP phase is transferred. In (46b), Neg is introduced into the derivation, constructing the NegP. In (46c), the nominative subject moves to Spec, TP. The accusative object moves into the TP domain, which does not violate anti-locality (the accusative object moves across the NegP). Thus, if NegP exists between vcanP and TP, the obligatory narrow scope interpretation of the accusative object may not be predicted by the present analysis. However, notice that many authors argue that Neg in Japanese is located within vP/VP (Takubo 1985; Han et al. 2004; Kataoka 2006). Building on these insights, I propose that Neg is an adjunct to vP. Given this proposal, when Neg and vcanP are merged, the resulting syntactic object is vcanP and not NegP:24

(47)

Returning to the obligatory narrow scope of the accusative object in (45a), the anti-locality analysis remains intact, as shown in (48):

(48)

Movement of the nominative subject and the accusative object:

In (48a), the object moves to the edge of the vcanP phase, and the vP phase is transferred. In (48b), Neg is introduced, constructing the vcanP (see (47b)). In (48c), the nominative subject moves into Spec, TP. The movement of the object violates anti-locality (the object moves across segments of the vcanP). The object must remain in its base-position and take scope under negation and the potential suffix:

(49)

The above analysis makes a further prediction, which is indeed borne out. Recall that the accusative object in the potential construction fails to move into the TP domain via scrambling because there are two phases (vcanP and vP), with transfer of the vP complement. The movement of the object to the edge of the vcanP, which is immediately below the TP, brings the object too close to the TP domain. Thus, the analysis predicts that accusative objects can move into the TP domain when they do not have to move to a phase edge immediately below the TP. Accusative objects in simple transitive sentences can take scope over negation (Kato 1985; Han et al. 2004; Kataoka 2006; Shibata 2015):

(50)

Kodomo-tati-ga

child-pl-nom

kanzirensyuu-dake-o

kanji.practice-only-nom

tuzuke-na-i.

continue-neg-prs

‘Children do not continue only kanji practice.’

‘Children do not continue kanji practice without continuing anything else.’ (not > only)

‘It is only kanji practice that children do not continue.’ (only > not)

The ambiguity of (50) is correctly predicted by the present analysis as transitive sentences involve only one phase (= the vP phase):

(51)

In (51a), both the subject and the object are base-generated within the vP phase. In (51b), Neg is introduced, constructing the vP (see (47b)). In (51c), the object and the nominative subject move into the TP domain (cf. Shibata 2015). The movement of the object does not violate anti-locality. Therefore, the object can take scope over negation.

The above analysis also predicts that the accusative object must take scope under negation when the former follows an instrumental subject:

(52)

Non-movement of the accusative object

 

a.

Constructing the vP phase:

                                                  [vP    pro    SUBJINS    [VP    OBJACC    V]    v]

 

b.

Merger of Neg:

                      [TP    proi    [vP    [vP      ti      SUBJINS    [VP    OBJACC    V]    v] Neg] T]

In (52), the instrumental subject, the pro, and the object are base-generated within the vP. In (52), Neg is introduced, constructing the vP. The instrumental subject remains within the vP and the pro moves to Spec, TP. The object remains within the VP, taking scope under negation. This prediction is borne out (Niinuma 2021):

(53)

Kodomo-tati-de

child-pl-with

kanzirensyuu-dake-o

kanji.practice-only-nom

tuzuke-na-i.

continue-neg-prs

‘Children do not continue only kanji practice.’

‘Children do not continue kanji practice without continuing anything else.’ (not > only)

‘It is only kanji practice that children do not continue.’ (?*only > not)

The accusative object in (53), which follows the instrumental subject, must take scope under negation.

 4.2 Apparent illicit movement of nominative phrases

I have argued above that the movement of VP-internal elements is constrained by phasal transfer and anti-locality (see (27) and (31)). In this subsection, I discuss some apparent counterexamples to this analysis and show that such cases in fact support it.

There do seem to be cases where nominative “objects” are in the TP domain even when an embedded object receives accusative Case (Tada 1992; 1993; Takano 2003).

(54)

a.

Kodomo-tati-ga

child-pl-nom

[kanzirensyuu-no

kanji.practice-gen

kiboosya]-o

applicant-acc

mituke-rare-ru.

find-can-prs

 

b.

Kodomo-tati-ga

child-pl-nom

kanzirensyuu-ga

kanji.practice-nom

kiboosya-o

applicant-acc

mituke-rare-ru.

find-can-prs

‘Children can find applicants for kanji practice.’

In (54a), kanzirensyuu ‘kanji practice’ receives genitive Case. Given that genitive Case is usually assigned to NP-internal elements (Kitagawa & Ross 1982), kanzirensyuu ‘kanji practice’ is within the NP headed by kiboosya ‘applicant’. In contrast, in (54b), kanzirensyuu ‘kanji practice’ receives nominative Case. Whereas (54a) can be analyzed as a case of the potential construction with the accusative object (26a), (54b) requires further discussion. The analysis developed above may predict that kanzirensyuu ‘kanji practice’ in (54b) cannot move into the TP domain:

(55)

Movement of the nominative object:

Here, kanzirensyuu ‘kanji practice’ in (55) moves out of the host NP headed by kiboosya ‘applicant’. However, there are several reasons to set aside this option. First, whether such a movement exists in Japanese is still debated, and even if it does, the alleged movement from the possessor position seems quite restricted (Funakoshi 2017). Moreover, if the NP under consideration could raise out of its host, the alleged movement into the TP domain should be blocked by anti-locality. As the extracted NP in (55) is at the outer edge of the vcanP, it is too close to the TP domain. However, if the NP does not move, then the subject cannot be accessed by T because of the condition on multiple phase edges in (20).

Thus, we are led to conclude that the nominative phrase in (54b) is base-generated within the TP region (cf. Saito 1982):

(56)

Non-movement of the nominative phrase:

In (56), the nominative subject moves to Spec, TP and the other nominative phrase is base-generated within the TP. The nominative phrase binds a pro associated with it.25 As the nominative phrase does not undergo movement, there is no anti-locality violation. This analysis is supported by several observations. First, the nominative phrase in (54b) obligatorily takes scope over negation:

(57)

a.

Kodomo-tati-ga

child-pl-nom

kanzirensyuu-dake-no

kanji.practice-only-gen

kiboosya-o

applicant-acc

mituke-rare-na-i.

find-can-neg-prs

‘Children cannot find applicants for only kanji practice.’ (not > only)

‘It is only kanji practice that children cannot find applicants for.’ (*only > not)

 

b.

Kodomo-tati-ga

child-pl-nom

kanzirensyuu-dake-ga

kanji.practice-only-nom

kiboosya-o

applicant-acc

mituke-rare-na-i.

find-can-neg-prs

‘Children cannot find applicants for only kanji practice.’ (*not > only)

‘It is only kanji practice that children cannot find applicants for.’ (only > not)

Whereas kanzirensyuu-dake ‘only kanji practice’ with genitive Case must take scope under negation in (57a), kanzirensyuu-dake ‘only kanji practice’ with nominative Case must take scope over negation in (57b). The contrast follows from the present analysis (recall that negation is an adjunct; see (47)):

(58)

The obligatory narrow scope interpretation of dake ‘only’ in (57a) follows because kanzirensyuu-dake-no ‘only kanji practice-gen’ is within the accusative NP that stays within the vP. The obligatory wide scope interpretation of dake ‘only’ in (57b) follows because the nominative phrase is base-generated above the potential suffix and negation in the TP domain. The absence of the narrow scope interpretation in (57) can be confirmed by the following example modeled after Nomura (2005) (see also Ochi & Isono 2021):

(59)

Kodomo-tati-ga

child-pl-nom

kanzirensyuu-dake-no/#ga

kanji.practice-only-gen/nom

kiboosya-o

applicant-acc

mituke-rare-ru

find-can-prs

no-wa

nmlz-top

sit-tei-ta

know-prog-pst

ga,

but

keesanrensyuu-dake-no

calculation.practice-only-gen

kiboosya-o

applicant-acc

mituke-rare-ru

find-can-prs

no-ni-wa

nmlz-dat-top

odoroi-ta.

be.surprised-pst

‘I have known that children can find applicants for only kanji practice, but I am surprised to

know that they can also find applicants for only calculation practice.’

Example (59) forces dake ‘only’ to take scope under the potential suffix. The context ensures that kanzirensyuu ‘kanji practice’ cannot be the only thing that the students can find applicants for. This example is a non-contradictory statement when kanzirensyuu-dake ‘kanji practice-only’ receives genitive Case. This means dake ‘only’ in this case can take scope under the potential suffix. In contrast, this example is clearly a contradictory statement when kanzirensyuu-dake ‘kanji practice-only’ receives nominative Case; dake ‘only’ in this case cannot take scope under the potential suffix. This contrast provides additional evidence for the structures in (58).26

 5. Conclusion

I have argued in this paper that the scope properties of Japanese objects are derived through an interaction between phasal transfer and anti-locality. The proposed analysis provides a unified account of the scope properties of objects and adjuncts in potential constructions. Additionally, assuming that Neg is an adjunct, the proposed analysis accounts for the scope properties of objects in simple transitive sentences. Furthermore, I discussed cases where nominative phrases apparently undergo illicit movement into the TP domain and showed that such cases can indeed be analyzed in terms of the base-generation of the nominative phrases, which provides additional evidence in favor of the proposed analysis. As the proposed analysis relies crucially on phasal transfer and a specific version of anti-locality, it offers new evidence in their favor.

 Abbreviations

acc = accusative, comp = complementizer, dat = dative, gen = genitive, ins = instrumental, neg = negation, nmlz = nominalizer, nom = nominative, pl = plural, prog = progressive, prs = present, pst = past, top = topic

The potential suffix is realized as -rare when the verb stem ends with a vowel and as -e when the stem ends with a consonant.

As noted by Koizumi (1994; 1998), accusative objects may take scope over the potential suffix when they are stressed. I discuss the scope properties of unstressed accusative objects in the main text; however, I return to the case of stressed accusative objects in footnote 21.

Note that dake ‘only’ in (4) precedes the postposition -de. However, dake ‘only’ can follow -de ‘with’, and in such cases, dake ‘only’ must take scope over the potential suffix (Kuno and Monane 1979; Shoji 1986; Futagi 2004; Shibata 2015, among others):

(i)

Kodomo-tati-ga

child-pl-nom

kanzirensyuu-o/ga

kanji.practice-acc/nom

pen-de-dake

pen-with-only

tuzuke-rare-ru.

continue-can-prs

(*can > only, only > can)

‘Children can continue kanji practice only with a pen.’

Dake ‘only’ follows -de ‘with’ in (i) and obligatorily takes scope over the potential suffix regardless of the Case of the object. I do not discuss this type of construction in this paper.

Instrumental subjects must be plural (Takubo 1984; Kishimoto 2005; 2010). The relevant examples in the text thus contain plural subjects.

The scope interpretation of the accusative object is not affected by subject marking:

(i)

Kodomo-tati-ga/de

child-pl-nom/with

kanzirensyuu-dake-o

kanji.practice-only-acc

tuzuke-rare-ru.

continue-can-prs

‘Children can continue only kanji practice.’

‘Children can continue kanji practice without continuing anything else.’ (can > only)

‘It is only kanji practice that children can continue.’ (?*only > can)

Here, the accusative object must take scope under the potential suffix with either kind of subject.

There are, in fact, many other approaches to the scope puzzle under consideration. Among them are the complex-head analysis (Saito & Hoshi 1998), the covert excorporation analysis (Saito 2012), and the prolepsis analysis (Takano 2003). Although they differ from one another in detail, all three analyses assume that nominative objects are always base-generated above the potential suffix. As they do not predict the availability of the narrow scope interpretation of the nominative object (5b), I set these analyses aside in the main text (see Nomura 2005; Bobaljik & Wurmbrand 2007; Takahashi 2010; 2011; 2021a for relevant discussion).

The accusative object may move to the (inner) Spec, vP (Tada 1992; 1993; Koizumi 1994; 1998; Nomura 2005). I set this movement aside as it is immaterial here.

I assume with Chomsky (2000; 2008; 2015) that the VP complement is transferred upon the completion of the vP phase for the sake of the exposition.

Note that vcanP does not project a phase (Nomura 2005), which means that the VP complement is not transferred upon the completion of vcanP. Hence, T can assign nominative Case to the object.

The narrow scope interpretation of the nominative object in (3b) may be obtained by reconstruction of the nominative object. Alternatively, the nominative object may stay in its base-generated position. See footnote 11.

I assume here that quantified elements do not need to move to a node of type t for scope interpretation.

I thank Michael Yoshitaka Erlewine for clarifying this point.

For Takahashi (2010; 2011), the domains of Case-valuation are phases.

As one reviewer correctly points out, the contrast in (15) is subsumed under the anaphor agreement effect (Rizzi 1990; Woolford 1999; Tucker 2012; Preminger 2019):

(i)

The anaphor agreement effect (Rizzi 1990: 26)

Anaphors do not occur in syntactic positions construed with agreement.

Whereas the subject in (15a) triggers subject agreement, the subject in (15b) does not. According to (i), the anaphor in the embedded subject position in (15a), but not that in (15b), is allowed. It would be interesting to see whether Saito’s (2017a; b; 2020) proposals can capture other cases subsumed by the anaphor agreement effect. I leave this question for future study.

I do not intend to claim that all types of movement must always target outer edges (see Wurmbrand 2013; Bošković 2016b; Longenbaugh 2019; Newman to appear, among others, for further discussion). Based on a paradigm concerning long-distance scrambling observed by Saito (1985), Takita (2010) argues that scrambling to vP must target the outer vP edge. I refer the reader to Takita (2010) for details.

Saito (2017a; b; 2020) argues that passive and unaccusative vs are also phases (see also Legate 2003 and Takahashi 2021b).

I argue below that objects and adjuncts following subjects in principle can undergo scrambling into the TP domain. Therefore, it is possible that the movement of the nominative object into the TP domain is an instance of scrambling rather than Case-movement. See Saito (2010a; 2012) and Takahashi (2010) for further discussion of the Case-movement analysis.

The contrast regarding the adjunct still holds even if the ordering of the adjunct and the object is reversed:

(i)

a.

Kodomo-tati-ga

child-pl-nom

pen-dake-de

pen-only-with

kanzirensyuu-o

kanji.practice-acc

tuzuke-rare-ru.

continue-can-prs

(can > only, ?*only > can)

 

b.

Kodomo-tati-ga

child-pl-nom

pen-dake-de

pen-only-with

kanzirensyuu-ga

kanji.practice-nom

tuzuke-rare-ru.

continue-can-prs

(?can > only, only > can)

‘Children can continue kanji practice with only a pen.’

The contrast in (i) is relevant to an interesting alternative analysis of the contrast in (30) suggested by one reviewer. The reviewer asks if the contrast in (30) can be accounted for in terms of movement into a Focus Phrase (FocP) between TP and vP (Hoshi & Miyoshi 2007). If elements containing dake ‘only’ move to FocP between TP and vP, the accusative object in (30a), which stays within the vP, blocks the movement of the adjunct, yielding the narrow scope interpretation of dake ‘only’ contained in the adjunct. However, the nominative object in (30b), which can move into the TP domain, does not block such movement and yields the wide scope interpretation of dake ‘only’ contained in the adjunct. Although the alternative may accommodate the contrast in (30), it cannot be extended to the contrast in (i). As the adjunct precedes the object, the latter should not interfere with the movement of the former. The contrast therefore remains puzzling under this alternative analysis.

Another reviewer asks if the contrast we observe in (i) holds for ditransitive constructions. As shown in (ii), the Case of direct objects impacts the scope of indirect objects when the former precede the latter:

(ii)

Scenario for the can > only reading: Taro, Hanako, John, and Mary are in a room. Taro can teach English only to Hanako after taking John and Mary to another room so that John and Mary do not have to learn English together.

Scenario for the only > can reading: Taro, Hanako, John, and Mary are in a room. Taro tries to teach English to Hanako, John, and Mary. However, only Hanako is willing to learn English, and neither John nor Mary listens to him.

 

a.

Taroo-ga

Taro-nom

eego-o

English-acc

Hanako-dake-ni

Hanako-only-dat

osie-rare-ru.

teach-can-prs

‘Taro can teach English to only Hanako.’

‘Taro can teach English to Hanako without teaching it to any other people.’ (can > only)

‘It is only Hanako that Taro can teach English to.’ (?*only > can)

 

b.

Taroo-ga

Taro-nom

eego-ga

English-nom

Hanako-dake-ni

Hanako-only-dat

osie-rare-ru.

teach-can-prs

‘Taro can teach English to only Hanako.’

‘Taro can teach English to Hanako without teaching it to any other people.’ (?can > only)

‘It is only Hanako that Taro can teach English to.’ (only > can)

The indirect object that contains dake ‘only’ fails to take scope over the potential suffix in the presence of the preceding accusative object (iia). In contrast, the indirect object takes scope over the potential suffix in the presence of the preceding nominative object (iib). Interestingly, the contrast seems to disappear or become significantly weaker when the indirect object precedes the direct object (I thank the reviewer for bringing this point to my attention):

(iii)

a.

Taroo-ga

Taro-nom

Hanako-dake-ni

Hanako-only-dat

eego-o

English-acc

osie-rare-ru.

teach-can-prs

‘Taro can teach English to only Hanako.’

‘Taro can teach English to Hanako without teaching it to any other people.’ (can > only)

‘It is only Hanako that Taro can teach English to.’ ((?)only > can)

 

b.

Taroo-ga

Taro-nom

Hanako-dake-ni

Hanako-only-dat

eego-ga

English-nom

osie-rare-ru.

teach-can-prs

‘Taro can teach English to only Hanako.’

‘Taro can teach English to Hanako without teaching it to any other people.’ (can > only)

‘It is only Hanako that Taro can teach English to.’ (only > can)

In (iiia) and (iiib), the indirect object can take scope over the potential suffix regardless of the Case of the direct object. As the structure of ditransitive constructions is still a matter of debate (Hoji 1985; Takano 2008, among others), I will not delve further into (iii) and will leave it for future investigation.

Note that the scrambling of the adjunct under consideration does not result in word order permutation. See footnote 21.

The scrambling of the object in (43b) targets CP, which is often assumed to be an A′-position. However, it is well-known that clause-internal scrambling exhibits characteristics of both A′-movement and A-movement (Saito 1992; Abe 1993; Tada 1993, among others) (I thank one reviewer for this point). I leave investigations into the properties of clause-internal scrambling for future study.

As Koizumi (1994; 1998) notes, the accusative object following the nominative subject may take scope over the potential suffix when the former is stressed, which Koizumi (1994; 1998) attributes to focus movement/scrambling. Note that in such a case, there is a pause after the accusative object:

(i)

Kodomo-tati-ga

child-pl-nom

kanzirensyuu-dake-o

kanji.practice-only-acc

[PAUSE]

 

tuzuke-rare-ru.

continue-can-prs

‘Children can continue only kanji practice.’

‘Children can continue kanji practice without continuing anything else.’ (can > only)

Of relevance here is Ko’s (2005a) observation that a subject and a following object may both undergo (string-vacuous) scrambling to a CP domain when there is a pause after the object (see also Miyagawa & Arikawa 2007). Following this insight, I suggest that (i) should be analyzed in terms of (string-vacuous) scrambling of the subject and the object to the CP domain:

(ii)

The subject and the object move into the CP domain from Spec, TP and Spec, vcanP, respectively. There is no anti-locality violation here, and the accusative object takes scope over the potential suffix. I assume here that clause-internal scrambling to the CP domain is not subject to radical (total) reconstruction (I thank one reviewer for clarifying this point). Note also that the instrumental subject and the following nominative object may take scope over the potential suffix when they are followed by a pause:

(iii)

Kodomo-tati-de

child-pl-with

kanzirensyuu-dake-ga

kanji.practice-only-nom

[PAUSE]

 

tuzuke-rare-ru.

continue-can-prs

‘Children can continue only kanji practice.’

‘Children can continue kanji practice without continuing anything else.’ (can > only)

‘It is only kanji practice that children can continue.’ (only > can)

(iv)

Kodomo-tati-dake-de

child-pl-only-with

[PAUSE]

 

kanzirensyuu-ga

kanji.practice-nom

tuzuke-rare-ru.

continue-can-prs

‘Only children can continue kanji practice.’

‘Children can continue kanji practice without any other people around.’ (can > only)

‘It is only children who can continue kanji practice.’ (only > not)

The nominative object in (iii) and the instrumental subject in (iv) can take scope over the potential suffix. I assume that the elements before the pause in (iii) and (iv) undergo scrambling to the CP domain (I thank one reviewer for suggesting that I consider such derivations). It is often claimed, or at least tacitly assumed, that such string-vacuous scrambling is impossible (Hoji 1985; Takita 2008). However, the status and empirical scope of the ban on string-vacuous scrambling do not seem to be clear at this point. Importantly, the double movement of the subject and object is indeed claimed to be possible (Ko 2005a; Miyagawa & Arikawa 2007; Shibata 2015; Sato & Maeda 2021). Hence, the analysis in the text is in line with the more recent understanding of string-vacuous scrambling, and I leave further investigations of the nature of the ban on string-vacuous scrambling for future study. Note that (33) and (51) also involve string-vacuous scrambling of the object and the adjunct.

One reviewer asks whether an analysis in terms of cyclic linearization (Fox & Pesetsky 2005; Ko 2005b; Takita 2008; 2010) can provide an analysis of the facts under consideration (I thank the reviewer for suggesting this alternative analysis). Although this analysis may accommodate some of the data discussed in the text, it is unclear how this analysis accounts for the obligatory narrow scope interpretation of the accusative object in the potential construction, which is the main focus of the current analysis. Under the alternative analysis, the relative order between elements established in a lower spell-out domain cannot be contradicted by the relative order established in a later spell-out domain. Specifically, this analysis allows elements contained in a lower spell-out domain to be accessible to operations after spell-out and undergo movement, as long as the movement does not yield ordering contradiction in a later spell-out domain. Therefore, elements contained in a spell-out domain do not have to move to the edge of the domain to be accessible to operations after the spell-out. The alternative analysis is thus silent about the obligatory narrow scope interpretation of the accusative object in the potential construction:

(i)

As shown in (ia), the object remains in the vP complement, which is spelled-out upon the completion of the vcanP phase. As the embedded PRO subject is phonetically empty, the embedded object is the only relevant phrase for establishing the relative word order. Hence, no relative order is created among phrases (V and v are set aside, as they always follow NPs owing to the head-finality of Japanese; see Takita 2010). Furthermore, in (ib), both the subject and the object can move into the TP domain without creating an ordering conflict or violating anti-locality. Nothing thus seems to block (i).

I thank one reviewer for his/her helpful comments that led to the reconsideration of the discussion in this section.

Interestingly, based on an analysis of subject-oriented floating quantifiers, Branan (2023) argues that Neg and some aspectual heads in Japanese should project NegP and AspP, respectively, above vP. These projections allow subject movement from Spec, vP to obey a version of anti-locality employed in Branan (2023). It is an interesting task to examine how we can reconcile the results of this paper with those obtained in Branan (2023), which I leave for future study.

One reviewer points out that the movement analysis in (56) should predict that a bound pronoun contained within the moved subject can be bound by the base-generated nominative phrase as a quantifier via reconstruction. However, as the reviewer correctly observes, such reconstruction is impossible (the following examples are provided by the reviewer (the judgment is the reviewer’s)):

(i)

a.

  Hotondo-no

  most-gen

sigotoi -ga

job-nom

[ sorei-o

  it-acc

yame-tagat-tei-ru

quit-want-prog-prs

hito-ga]j

person-nom

koonin-o

replacement-acc

mituke-rare-na-i.

find-can-neg-prs

 

b.

*[Sorei-o

    it-acc

yame-tagat-tei-ru

quit-want-prog-prs

hito-ga]j

person-nom

hotondo-no

most-gen

sigotoi -ga

job-nom

tj

 

koonin-o

replacement-acc

mituke-rare-na-i.

find-can-neg-prs

   ‘Most of the jobs are such that those who want to quit them cannot find a replacement.’

When the base-generated nominative phrase hotondo-no sigoto ‘most of the jobs’ precedes the moved nominative phrase sore-o yame-tagatteiru hito ‘those who want to quit them’, as in (ia), sore ‘it’ can be bound by hotondo-no sigoto ‘most of the jobs’. Interestingly, when the ordering of the two nominative phrases is reversed, as in (ib), sore ‘it’ cannot be bound by hotondo-no sigoto ‘most of the jobs’. The unacceptability of (ib) is puzzling given that pronominal variable binding is claimed to be possible via reconstruction (Fox 2000); the movement analysis seems to predict that (ib) should be as acceptable as (ia). However, the contrast between (ia) and (ib) does not undermine the movement analysis proposed in the text. First, I assume that in (ia), the nominative phrase hotondo-no sigoto ‘most of the jobs’ is base-generated above the moved nominative phrase, which makes the pronominal variable binding in question possible. Second, nominative phrases in potential constructions in general do not seem to permit reconstruction for pronominal variable binding (see Takano 2003 for relevant discussion), which makes the unacceptability of (ib) consistent with the movement analysis. Pronouns contained in objects can be bound by quantificational subjects when the former follow the latter regardless of the Case of the latter:

(ii)

a.

Hotondo-no

most-gen

kigyooi-ga

company-nom

sokoi-no

it-gen

danseezyuugyooin-o

male.employee-acc

suguni

immediately

kaiko-deki-ru.

fire-can-prs

 

b.

Hotondo-no

most-gen

kigyooi-ni(-wa)

company-dat-top

sokoi-no

it-gen

danseezyuugyooin-ga

male.employee-nom

suguni

immediately

kaiko-deki-ru.

fire-can-prs

‘Most companies can fire their male employees immediately.’

The pronoun soko ‘it’, contained within the accusative object in (iia) and the nominative object in (iib), can be bound by the nominative subject and the dative subject, respectively. Importantly, when the objects in (ii) are moved to sentence-initial position, pronominal variable binding is not available with the nominative object:

(iii)

a.

    Sokoi-no

    it-gen

danseezyuugyooinj-o

male.employee-acc

hotondo-no

most-gen

kigyooi-ga

company-nom

tj

 

suguni

immediately

kaiko-deki-ru.

fire-can-prs

 

b.

?*Sokoi-no

    it-gen

danseezyuugyooinj-ga

male.employee-nom

hotondo-no

most-gen

kigyooi-ni(-wa)

company-dat-top

tj

 

suguni

immediately

kaiko-deki-ru.

fire-can-prs

    ‘Most companies can fire their male employees immediately.’

While the pronoun sore ‘it’ within the scrambled accusative object can be bound by the nominative subject, as shown in (iiia), the pronoun sore ‘it’ within the scrambled nominative object fails to be bound by the dative subject, as shown in (iiib). Given that the moved nominative object in (iiib) does not reconstruct for pronominal variable binding, it is not surprising that the moved subject in (ib) also does not reconstruct for pronominal variable binding. I leave investigations of this restriction for future study.

Note that the nominative “object” can take scope over the potential suffix when it precedes the instrumental subject:

(i)

Kanzirensyuu-dake-ga

kanji.practice-only-nom

kodomo-tati-de

child-pl-with

tuzuke-rare-ru.

continue-can-prs

‘Children can continue only kanji practice.’

‘Children can continue kanji practice without continuing anything else.’ (can > only)

‘It is only kanji practice that children can continue.’ (only > can)                                                                          (Takahashi 2021a: 162)

Kanzirensyuu-dake ‘only kanji practice’ is in sentence-initial position and can take scope over the potential suffix. Given that nominative phrases can be base-generated in the TP domain, (i) may indeed be structurally ambiguous: (i) may involve movement of kanzirensyuu-dake ‘only kanji practice’ to the TP domain or base-generation of kanzirensyuu-dake ‘only kanji practice’ within the TP domain (irrelevant parts are omitted below).

(ii)

a.

b.

[TP

[TP

NPiNOM

NPiNOM

[vcanP

[vcanP

[vP

[vP

SUBJINS

SUBJINS

[VP

[VP

   ti

proi

V] tj] vj-vcan] T] (= (i))

V] tj] vj-vcan] T] (= (i))

Anti-locality is not violated by either option. I leave further investigations of this point for another occasion. The analysis in the text also predicts that when the instrumental subject undergoes A′-movement into a CP domain, the following nominative phrase may take scope over the potential suffix (I thank one reviewer for suggesting that I consider such cases; see also footnote 21):

(iii)

[CP    SUBJiINS    [TP    NPjNOM    [vcanP    [vP    ti [VP    proj/tj    V] tk]    vk-vcan] T] C]

In (iii), the instrumental subject moves to Spec, CP, and the following nominative phrase, which may be base-generated in the TP domain or move into the TP domain, is located above the potential suffix. This prediction is borne out:

(iv)

Kodomo-tati-de-wa

child-pl-with-top

kanzirensyuu-dake-ga

kanji.practice-only-nom

tuzuke-rare-ru.

continue-can-prs

‘Children can continue only kanji practice.’

‘Children can continue kanji practice without continuing anything else.’ (can > only)

‘It is only kanji practice that children can continue.’ (only > can)

The instrumental marker -de ‘with’ is a postposition. Given that topicalization of PPs involves movement into the CP domain (Saito 2010b), the availability of the wide scope interpretation of the nominative object in (iv) bears out the prediction shown in (iii). Example (v), modified from the example given by the reviewer, also illustrates this point:

(v)

Rokunensee-de-moi

sixth.graders-with-even

sensee-wa

teacher-top

[CP

 

ti

 

kanzirensyuu-dake-ga

kanji.practice-only-nom

tuzuke-rare-ru

continue-can-prs

to]

comp

omot-ta.

think-pst

‘Lit. Even the sixth-gradersi, the teacher thinks that ti can continue only kanji practice.’

‘The teacher thinks that even the sixth-graders can continue kanji practice without continuing anything else.’ (can > only)

‘The teacher thinks that it is only kanji practice that even the sixth-graders can continue.’ (only > can)

In (v), the instrumental subject moves out of the complement clause, and the embedded nominative phrase can take scope either over or under the potential suffix. As the nominative phrase can be located within the embedded TP via either base-generation or movement, the former can take scope over the potential suffix.

 Funding information

This work was partially supported by JSPS KAKENHI Grant Number JP 20K00535.

 Acknowledgements

I am very grateful to four anonymous reviewers for Glossa and Michael Yoshitaka Erlewine for their extremely helpful suggestions on the earlier versions of this paper. I would also like to thank Sato Ebina, Sana Niinuma, Hisako Takahashi, Naoto Tomizawa, and the participants in the Workshop on Comparative Syntax and Language Acquisition (CSLA) #11 (online), hosted by Center for Linguistics, Nanzan University, for helpful comments and/or discussions. Any remaining errors are my own.

 Competing interests

The author has no competing interests to declare.

 Abe, Jun. 1993. Binding conditions and scrambling without A/A′ distinctions. Storrs, CT: University of Connecticut dissertation. Abels, Klaus. 2003. Successive cyclicity, anti-locality, and adposition stranding. Storrs, CT: University of Connecticut dissertation. Bobaljik, Jonathan David & Wurmbrand, Susi. 2005. The domain of agreement. Natural Language & Linguistic Theory 23(4). 809–865. DOI: 10.1007/s11049-004-3792-4 Bobaljik, Jonathan David & Wurmbrand, Susi. 2007. Complex predicates, aspect, and anti-reconstruction. Journal of East Asian Linguistics 16(1). 27–42. DOI: 10.1007/s10831-006-9004-y Boeckx, Cedric & Grohmann, Kleanthes K. 2007. Remark: Putting phases in perspective. Syntax 10(2). 204–222. DOI: 10.1111/j.1467-9612.2007.00098.x Bošković, Željko. 1994. D-structure, theta-criterion, and movement into theta-positions. Linguistic Analysis 24 (3–4). 247–286. Bošković, Željko. 1997. The syntax of nonfinite complementation: An economy approach. Cambridge, MA: MIT Press. Bošković, Željko. 2005. On the locality of left branch extraction and the structure of NP. Studia Linguistica 59(1). 1–45. DOI: 10.1111/j.1467-9582.2005.00118.x Bošković, Željko. 2007. On the locality and motivation of Move and Agree: An even more minimal theory. Linguistic Inquiry 38(4). 589–644. DOI: 10.1162/ling.2007.38.4.589 Bošković, Željko. 2014a. Now I’m a phase, now I’m not a phase: On the variability of phases with extraction and ellipsis. Linguistic Inquiry 45(1). 27–89. DOI: 10.1162/LING_a_00148 Bošković, Željko. 2014b. More on the edge of the edge. In Chapman, Cassandra & Kit, Olena & Kučerová, Ivona (eds.), Formal Approaches to Slavic Linguistics 22, 44–46. Ann Arbor, MI: Michigan Slavic Publications. Bošković, Željko. 2015. From the Complex NP Constraint to everything: On deep extractions across categories. The Linguistic Review 32(4). 603–669. DOI: 10.1515/tlr-2015-0006 Bošković, Željko. 2016a. What is sent to spell-out is phases, not phasal complements. Linguistica 56(1). 25–66. DOI: 10.4312/linguistica.56.1.25-66 Bošković, Željko. 2016b. Getting really edgy: On the edge of the edge. Linguistic Inquiry 47(1). 1–33. DOI: 10.1162/LING_a_00203 Branan, Kenyon. 2023. Locality and antilocality: The logic of conflicting requirements. Linguistic Inquiry 51(1). 1–38. DOI: 10.1162/ling_a_00436 Chomsky, Noam. 2000. Minimalist inquiries: The framework. In Martin, Roger & Michaels, David & Uriagereka, Juan (eds.), Step by step: Essays on minimalist syntax in honor of Howard Lasnik, 89–155. Cambridge, MA: MIT Press. Chomsky, Noam. 2008. On phases. In Freidin, Robert & P. Otero, Carlos & Zubizarreta, Maria Luisa (eds.), Foundational issues in linguistic theory, 133–166. Cambridge, MA: MIT Press. DOI: 10.7551/mitpress/7713.003.0009 Chomsky, Noam. 2015. Problems of projection: Extensions. In Di Domenico, Elisa & Hamann, Cornelia & Matteini, Simona (eds.), Structures, strategies and beyond: Studies in honour of Adriana Belletti, 3–16. Amsterdam/Philadelphia: John Benjamins. DOI: 10.1075/la.223.01cho Ebina, Sato. 2020. Kanoozyutugobun-ni okeru syukakumokutekigo-no sayooiki-ni tuite no bunseki [An analysis of the scope of nominative objects in potential constructions]. B.A. thesis, Yamagata University. Epstein, Samuel D. & Kitahara, Hisatsugu & Seely, Daniel. 2016. Phase cancellation by external pair-merge of heads. The Linguistic Review 33(1). 87–102. DOI: 10.1515/tlr-2015-0015 Erlewine, Michael Yoshitaka. 2020. Anti-locality and subject extraction. Glossa 5(1): 84. 1–38. DOI: 10.5334/gjgl.1079 Fukui, Naoki. 1986. A theory of category projection and its applications. Cambridge, MA: Massachusetts Institute of Technology dissertation. Fox, Danny. 2000. Economy and semantic interpretation. Cambridge, MA: MIT Press. Fox, Danny & Pesetsky, David. 2005. Cyclic linearization and its interaction with other aspects of the grammar. Theoretical Linguistics 31. 235–262. DOI: 10.1515/thli.2005.31.1-2.235 Funakoshi, Kenshi. 2017. Backward control from possessors. Syntax 20(2). 170–213. DOI: 10.1111/synt.12134 Funakoshi, Kenshi & Takahashi, Masahiko. 2014. LF intervention effects and nominative objects in Japanese. In Kwon, Soohyun (ed.), Proceedings of the 37th annual Penn Linguistics Conference. 101–110. University of Pennsylvania Working Papers in Linguistics. Funakoshi, Sayaka. 2015. A theory of generalized pied-piping. College Park, MD: University of Maryland dissertation. Futagi, Yoko. 2004. Japanese focus particles at the syntax-semantics interface. New Brunswick, NJ: Rutgers University dissertation. Grohmann, Kleanthes K. 2003. Prolific domains: On the anti-locality of movement dependencies. Amsterdam/Philadelphia: John Benjamins. DOI: 10.1075/la.66 Han, Chung-Hye & Storoshenko, Dennis Ryan & Sakurai, Yasuko. 2004. Scope of negation and clause structure in Japanese. In Ettlinger, Marc & Fleisher, Nicholas & Park-Doob, Mischa (eds.), Proceedings of the 30th Berkeley linguistics society, 118–129. University of California, Berkley, CA: Berkley Linguistics Society. DOI: 10.3765/bls.v30i1.955 Hiraiwa, Ken. 2001. Multiple Agree and the defective intervention constraint in Japanese. In Lance, Nathan & Costa, Albert & Martin-Gonzalez, Javier & Matushansky, Ora & Szczegielniak, Adam (eds.), Proceedings of the 1st HUMIT student conference in language research (HUMIT 2000), 67–80. Cambridge, MA: MITWPL. Hoji, Hajime. 1985. Logical Form constraints and configurational structure in Japanese. Seattle, WA: University of Washington dissertation. Holmberg, Anders & Sheehan, Michelle & van der Wal, Jenneke. 2019. Movement from the double object construction is not fully symmetrical. Linguistic Inquiry 50(4). 677–722. DOI: 10.1162/ling_a_00322 Hoshi, Koji & Miyoshi, Nobuhiro. 2007. A derivational approach to dake: A preliminary sketch of association with focus. Language, Culture, and Communication 39. 25–48. Inoue, Kazuko. 1998. Sentences without nominative subjects in Japanese. In Grant-in-Aid for COE research report (2A): Researching and verifying an advanced theory of human language, 1–34. Chiba: Kanda University of International Studies. Ishii, Toru. 1999. Cyclic spell-out and the that-t effects. In Bird, Sonya & Carnie, Andrew & Haugen, Jason D. & Norquest, Peter (eds.), Proceedings of West Coast Conference on Formal Linguistics 18, 220–231. Somerville, MA: Cascadilla Press. Kasai, Hironobu. 2018. Case valuation after scrambling: Nominative objects in Japanese. Glossa 3(1): 127. 1–29. DOI: 10.5334/gjgl.676 Kataoka, Kiyoko. 2006. Nihongo hiteibun-no koozoo: Kakimazebun-to hiteikooo-hyoogen [Syntactic structure of Japanese negative sentences: Scrambling construction and negation-sensitive elements]. Tokyo: Kuroshio Publishers. Kato, Yasuhiko. 1985. Negative sentences in Japanese. Tokyo: Sophia University. Kishimoto, Hideki. 2005. Toogo koozoo-to bunpoo kankei [Syntactic structures and grammatical relations]. Tokyo: Kurosio Publishers. Kishimoto, Hideki. 2010. Subjects and constituent structure in Japanese. Linguistics 48(3). 629–670. DOI: 10.1515/ling.2010.020 Kishimoto, Hideki. 2017. Negative polarity, A-movement, and clause architecture in Japanese. Journal of East Asian Linguistics 26. 109–161. DOI: 10.1007/s10831-016-9153-6 Kitagawa, Chisato & Ross, Claudia N. G. 1982. Prenominal modification in Chinese and Japanese. Linguistic Analysis 9. 19–53. Ko, Heejeong. 2005a. Syntax of why-in-situ: Merge into [Spec, CP] in the overt syntax. Natural Language and Linguistic Theory 23. 867–916. DOI: 10.1007/s11049-004-5923-3 Ko, Heejeong. 2005b. Syntactic edges and linearization. Cambridge, MA: Massachusetts Institute of Technology dissertation. Koizumi, Masatoshi. 1994. Nominative objects: The role of TP in Japanese. In Koizumi, Masatoshi & Ura, Hiroyuki (eds.), Proceedings of the first formal approaches to Japanese linguistics (FAJL1), 211–230. Cambridge, MA: MITWPL. Koizumi, Masatoshi. 1998. Remarks on nominative objects. Journal of Japanese Linguistics 16. 39–66. DOI: 10.1515/jjl-1998-0104 Kuno, Susumu. 1973. The structure of the Japanese language. Cambridge, MA: MIT Press. Kuno, Susumu & Ajiro Monane, Tazuko. 1979. Positioning of quantifier-like particles. The Journal of the Association of Teachers of Japanese 14(2). 115–140. DOI: 10.2307/489041 Kuroda, Shige-Yuki. 1988. Whether we agree or not: A comparative syntax of English and Japanese. Lingvisticae Investigationes 12(1). 1–47. DOI: 10.1075/li.12.1.02kur Legate, Julie Anne. 2003. Some interface properties of the phase. Linguistic Inquiry 34(3). 506–516. DOI: 10.1162/ling.2003.34.3.506 Longenbaugh, Nicholas. 2019. On expletives and the agreement/movement correlation. Cambridge, MA: Massachusetts Institute of Technology dissertation. Miyagawa, Shigeru & Arikawa, Koji. 2007. Locality in syntax and floating numeral quantifiers. Linguistic Inquiry 38(4). 645–670. DOI: 10.1162/ling.2007.38.4.645 Moritake, Nozomi. 2022. A focus-based analysis of nominative objects in Japanese. Lingua 278. 1–31. DOI: 10.1016/j.lingua.2022.103404 Newman, Elise. to appear. The order of operations and A/A-bar interactions. Natural Language and Linguistic Theory. Niinuma, Sana. 2021. Nihongo ni okeru hiteizi to mokutekigo no sayooikikankei ni kansuru kenkyu [A study on the scope relation between negation and objects in Japanese]. B.A. thesis, Yamagata University. Nomura, Masashi. 2005. Nominative Case and AGREE(ment). Storrs, CT: University of Connecticut dissertation. Nomura, Masashi. 2020. Labeling and pair-merge of heads. Nanzan Linguistics 15. 1–24. Ochi, Masao & Isono, Asuka. 2021. Agree, move, and nominative objects in Japanese. Nanzan Linguistics 16. 81–106. Preminger, Omer. 2019. The Anaphor Agreement Effect: Further evidence against binding-as-agreement. Ms., University of Maryland, College Park. Quicoli, A. Carlos. 2008. Anaphora by phase. Syntax 11(3). 299–329. DOI: 10.1111/j.1467-9612.2008.00116.x Richards, Norvin. 2001. Movement in language: Interactions and architectures. Oxford: Oxford University Press. DOI: 10.1093/oso/9780198241171.001.0001 Rizzi, Luigi. 1990. On the Anaphor Agreement Effect. Rivista di Grammatica Generativa 2. 27–42. Saito, Mamoru. 1982. Case marking in Japanese: A preliminary study. Ms., Massachusetts Institute of Technology. Saito, Mamoru. 1985. Some asymmetries in Japanese and their theoretical implications. Cambridge, MA: Massachusetts Institute of Technology dissertation. Saito, Mamoru. 1989. Scrambling as semantically vacuous A-movement. In Baltin, Mark R. & Kroch, Anthony S. (eds.), Alternative conceptions of phrase structure, 189–200. Chicago, Illinois: The University of Chicago Press. Saito, Mamoru. 1992. Long distance scrambling in Japanese. Journal of East Asian Linguistics 1. 69–118. DOI: 10.1007/BF00129574 Saito, Mamoru. 2006. Subjects of complex predicates: A preliminary study. In Kawamura, Tomoko & Suh, Yunju, & Larson, Richard K. (eds.), Stony Brook occasional papers in linguistics 1. 172–188. Department of Linguistics, Stony Brook University. Saito, Mamoru. 2010a. On the scope properties of nominative phrases in Japanese. In Mohanty, Rajat & Menon, Mythili (eds.), Universals and variation: Proceedings of GLOW in Asia VII 2009, 313–333. Hyderabad: EFL University Press. Saito, Mamoru. 2010b. Semantic and discourse interpretation of the Japanese left periphery. In Erteschik-Shir, Nomi & Rochman, Lisa (eds.), The sound patterns of syntax, 140–173. Oxford: Oxford University Press. DOI: 10.1093/acprof:oso/9780199556861.003.0008 Saito, Mamoru. 2012. Case checking/valuation in Japanese: Move, Agree or Merge? Nanzan Linguistics 8. 109–127. Saito, Mamoru. 2017a. A note on transfer domains. Nanzan Linguistics 12. 61–69. Saito, Mamoru. 2017b. Notes on the locality of anaphor binding and A-movement. English Linguistics 34(1). 1–33. DOI: 10.9793/elsj.34.1_1 Saito, Mamoru. 2017c. Labeling and argument doubling in Japanese. Tsing Hua Journal of Chinese Studies 47(2). 383–405. DOI: 10.6503/THJCS.2017.47(2).07 Saito, Mamoru. 2018. Kase as a weak head. In Kalin, Laura & Paul, Ileana & Vander Klok, Jozina (eds.), McGill Working Papers in Linguistics 25 (1): Heading in the right direction: Linguistic treats for Lisa Travis, 382–190. McGill Working Papers in Linguistics. Saito, Mamoru. 2020. On the causative paradoxes: Derivations and transfer domains. Nanzan Linguistics 15. 25–44. Saito, Mamoru & Hoshi, Hiroto. 1998. Control in complex predicates. In Report of the special research project for the typological investigation of languages and cultures of the East and West, 15–46. University of Tsukuba. Sano, Masaki. 1985. LF movement in Japanese. Descriptive and Applied Linguistics 18. 245–259. Sato, Yosuke & Maeda, Masako. 2021. Syntactic head movement in Japanese: Evidence from verb-echo answers and negative scope reversal. Linguistic Inquiry 52(2). 359–376. DOI: 10.1162/ling_a_00380 Sheehan, Michelle & Cyrino, Sonia. To appear. Restrictions on long passives in English and Brazilian Portuguese: A phase-based account. Linguistic Inquiry. DOI: 10.1162/ling_a_00482 Shibata, Yoshiyuki. 2015. Exploring syntax from the interfaces. Storrs, CT: University of Connecticut dissertation. Shimamura, Koji & Wurmbrand, Susi. 2014. Two types of restructuring in Japanese: Evidence from scope and binding. In Kawahara, Shigeto & Igarashi, Mika (eds.), Proceedings of FAJL 7: Formal approaches to Japanese linguistics, 203–214. Cambridge, MA: MITWPL. Shoji, Atsuko. 1986. Dake and sika in Japanese: Syntax, semantics and pragmatics. Ithaca, NY: Cornell University dissertation. Tada, Hiroaki. 1992. Nominative objects in Japanese. Journal of Japanese Linguistics 14. 91–108. DOI: 10.1515/jjl-1992-0105 Tada, Hiroaki. 1993. A/A-bar partition in derivation. Cambridge, MA: Massachusetts Institute of Technology dissertation. Takahashi, Masahiko. 2010. Case, phases, and nominative/accusative conversion in Japanese. Journal of East Asian Linguistics 19. 319–355. DOI: 10.1007/s10831-010-9063-y Takahashi, Masahiko. 2011. Some theoretical consequences of Case-marking in Japanese. Storrs, CT: University of Connecticut dissertation. Takahashi, Masahiko. 2021a. Some notes on the scope properties of nominative objects in Japanese. In Laszakovits, Sabine & Shen, Zheng (eds.), The size of things I: Structure building, 153–171. Berlin: Language Science Press. https://zenodo.org/records/5524288 Takahashi, Masahiko. 2021b. Reconsidering the optionality of raising in Japanese exceptional Case-marking constructions. Syntax 24(2). 224–262. DOI: 10.1111/synt.12206 Takano, Yuji. 2003. Nominative objects in Japanese complex predicate constructions: A prolepsis analysis. Natural Language and Linguistic Theory 21(4). 779–834. DOI: 10.1023/A:1025545313178 Takano, Yuji. 2008. Ditransitive constructions. In Miyagawa, Shigeru & Saito, Mamoru (eds.), The Oxford handbook of Japanese linguistics, 423–455. New York: Oxford University Press. DOI: 10.1093/oxfordhb/9780195307344.013.0016 Takano, Yuji. 2010. Scrambling and control. Linguistic Inquiry 41(1). 83–110. DOI: 10.1162/ling.2010.41.1.83 Takezawa, Koichi. 1987. A configurational approach to Case-marking in Japanese. Seattle, WA. University of Washington dissertation. Takita, Kensuke. 2008. String-vacuous scrambling and cyclic linearization. In Otsu, Yukio (ed.), The proceedings of the 9th Tokyo Conference on Psycholinguistics (TCP 2008), 225–249. Tokyo: Hituzi Syobo. Takita, Kensuke. 2010. Cyclic linearization and constraints on movement and ellipsis. Nagoya: Nanzan University dissertation. Takubo, Yukinori. 1984. Gendai nihongo-no basyo-o arawasu meisirui-ni-tuite [Nouns indicating places in Modern Japanese]. Nihongo-nihonbunka 12. 89–117. Osaka University of Foreign Studies. Takubo, Yukinori. 1985. On the scope of negation in Japanese. Papers in Japanese Linguistics 10. 87–115. DOI: 10.1515/jjl-1985-1-208 Tucker, Matthew. 2012. On the derivation of the Anaphor Agreement Effect. Unpublished Ms., UC Santa Cruz. Ura, Hiroyuki. 1999. Checking theory and dative subject constructions in Japanese and Korean. Journal of East Asian Linguistics 7. 223–254. DOI: 10.1023/A:1008383332362 Uriagereka, Juan. 1999. Multiple spell-out. In Epstein, Samuel & Hornstein, Norbert (eds.), Working minimalism, 251–282. Cambridge, MA: MIT Press. Woolford, Ellen. 1999. More on the Anaphor Agreement Effect. Linguistic Inquiry 30(2). 257–287. DOI: 10.1162/002438999554057 Wurmbrand, Susi. 2001. Infinitives: Restructuring and clause structure. Berlin/New York: De Gruyter Mouton. Wurmbrand, Susi. 2013. Tagalog infinitives: Consequences for the theory of phases, voice marking and extraction. LingBuzz 001898. Yang, Dong-Whee. 1983. The extended binding theory of anaphors. Language Research 19. 169–192.