1 Introduction

Javanese (ISO 639-3 jav), an Austronesian language of Java, Indonesia, exhibits a di-marked construction traditionally described as a passive (Wedhawati & Arifin 2006; Robson 2014; Krauße 2017). This construction (1a) is characterized by an optional agent by-phrase and a theme argument in preverbal position—a position restricted to the syntactically prominent phrase in the clause. It also involves dedicated morphological marking (di-), which is absent in its active counterpart (1b), as is typical of passive constructions cross-linguistically (Shibatani 1985).

    1. (1)
    1. a.
    1. Joko
    2. Joko
    1. wis
    2. already
    1. ḍi-gènḍong
    2. pass-carry
    1. (ambè’
    2. (by
    1. Bambang).
    2. Bambang)
    1. ‘Joko was carried (by Bambang).’(Javanese)
    1.  
    1. b.
    1. Bambang
    2. Bambang
    1. wis
    2. already
    1. ng-gènḍong
    2. av-carry
    1. Joko.
    2. Joko
    1. ‘Bambang has already carried Joko.’

This construction shares key typological properties with passive constructions in neighboring Austronesian languages, such as Indonesian (ISO 639-3 ind) and Balinese (ISO 639-3 ban). Both are widely analyzed as exhibiting true passives involving A-movement of the theme to subject position (Wechsler & Arka 1998; Aldridge 2008; Arka 2008; Cole et al. 2008; Kroeger 2014; inter alia). The Indonesian passive (2) features the same verbal affix di- as the Javanese construction in (1); the Balinese passive (3) employs a different affix, but displays the same word order pattern.1

    1. (2)
    1. Kue
    2. cake
    1. ini
    2. this
    1. di-makan
    2. pass-eat
    1. (oleh
    2. (by
    1. Arna).
    2. Arna)
    1. ‘This cake was eaten (by Arna).’ (Cole et al. 2008: 1509) (Standard Indonesian)
    1. (3)
    1. Nasi-ne
    2. rice-def
    1. ajeng-a
    2. eat-pass
    1. (teken
    2. (by
    1. anak-e
    2. person-def
    1. ento).
    2. that)
    1. ‘The rice was eaten by that person.’ (Artawa 1998: 10)(Balinese)

While these observations suggest that the Javanese di-construction is a typical Indonesian-type passive, new data from East Javanese, a lesser-studied dialect of eastern Java (including the major cities of Surabaya and Malang; Ras 1994; Krauße 2017), reveal a puzzling asymmetry that calls this analysis into question. In East Javanese actor voice clauses, a numeral quantifier (e.g. rolas ‘twelve’) can freely dislocate from its clause-initial host and intervene between the host and an aspectual auxiliary, such as the future auxiliary até in (4).

    1. (4)
    1. [___i
    2. [___
    1. Konco-ku]
    2. friend-1sg]
    1. rolasi
    2. twelve
    1. até
    2. fut
    1. mangan
    2. av.eat
    1. tahu.
    2. tofu
    1. ‘My twelve friends will eat tofu.’ (actor voice)

When a clause is marked in the so-called passive voice, quantifier float in the same pre-auxiliary field becomes unacceptable. This is shown in (5), where dislocating the same numeral quantifier rolas ‘twelve’ between the sentence-initial theme (e.g. ‘the tofu’) and the future auxiliary até yields semantic and grammatical consequences.

    1. (5)
    1. *[___i
    2.   [___
    1. Tahu-né]
    2. tofu-def]
    1. rolasi
    2. twelve
    1. até
    2. fut
    1. ḍi-pangan
    2. 3/pass-eat
    1. (ambè’
    2. (by
    1. konco-ku).
    2. friend-1sg)
    1.   (intended: ‘The twelve pieces of tofu will be eaten (by my friend).’) (passive voice)

This voice-based asymmetry is unexpected if the di-construction (5) is a true passive. Under a widely adopted formal approach to Indonesian-type Austronesian languages (e.g. Aldridge 2008; Cole et al. 2008; Legate 2014; inter alia), actor voice (AV) exhibits nominative-accusative alignment, with T assigning nominative case to the highest caseless DP, as illustrated in (6a). The passive is analyzed as a subtype of AV, in which the theme A-moves to [Spec TP] to receive nominative case (6b), parallel to the external argument in transitive AV clauses (6a).

    1. (6)
    1. The traditional A-approach to Indonesian-type AV-passive alternation2
    1.  
    1. a.
    1. Actor Voice
    1.  
    1. b.
    1. Passive Voice

If this analysis holds for East Javanese, no asymmetry in quantifier float is expected between AV and the passive, as both involve the same A-movement step from the VoiceP phase edge to the subject position [Spec TP]. The availability of the pre-auxiliary landing site in AV but not in the passive, as indicated by the quantifier float asymmetry between (4) and (5), therefore remains unaccounted for under this analysis.

Building on this voice-based asymmetry, this paper pursues three goals. First, it demonstrates that the East Javanese di-construction is distinct from Indonesian-type passives (e.g. (2)–(3)) and involves no A-movement of the theme to the grammatical subject position. Second, it argues that this pseudo-passive construction should be unified with the object voice (OV) construction (7), which alternates with actor voice and together forms what is commonly referred to as an Indonesian-type voice system. Like the di-construction, the OV construction features a preverbal theme argument that is disallowed in actor voice, as shown below in (7).

    1. (7)
    1. East Javanese object voice
    1. Arè’
    2. child
    1. iku
    2. dem
    1. wis
    2. perf
    1. ta’=Ø-rangkul.
    2. 1sg=ov-hug
    1. ‘I hugged that child.’

Drawing on six diagnostics, we present novel evidence that the East Javanese OV construction, of which the di-construction is a subtype, is best analyzed as an object topicalization construction. On this view, the preverbal theme in (7) is a topic phrase occupying an Ā-position, as schematized in (8a). We then show that the East Javanese AV construction contrasts with OV in involving subject topicalization, as illustrated in (8b): the highest caseless DP first moves to the grammatical subject position ([Spec TP] in (8b)) and subsequently undergoes Ā-movement to a left-peripheral topic position. On this analysis, the contrast between the di-construction (i.e. an object topicalization construction) and AV in disallowing quantifier float in the same pre-auxiliary field (i.e. the presumed subject position, [Spec TP]) (4) follows straightforwardly.

    1. (8)
    1. The proposed Ā-approach to the East Javanese voice system
    1.  
    1. a.
    1. OV/“passive” (object topicalization construction)
    1.  
    1. b.
    1. AV (subject topicalization construction)

The final goal of this paper is to examine how the East Javanese pseudo-passive sheds light on the developmental pathways of Indonesian-type passive constructions and, more broadly, on the evolution of syntax in Western Austronesian languages. Specifically, the coexistence of the Javanese pseudo-passive with morphologically identical di-marked passives in neighboring languages points to a plausible pathway whereby topicalization is reanalyzed as passivization through the grammaticalization of topics into subjects, in line with observations from other language families (e.g. Comrie 1988; Shibatani 2011). This account further addresses a longstanding question in Austronesian syntax concerning how true passive constructions emerged from symmetrical voice systems.

The remainder of the paper is structured as follows. Section 2 introduces the key properties of the East Javanese voice system. Section 3 presents evidence from six diagnostics that both the di-construction and the so-called object voice are best analyzed as instances of object topicalization. Section 4 compares the East Javanese system with those found in other Javanese dialects and Indonesian-type Austronesian languages, highlighting its relative conservatism, an underexplored dimension of syntactic variation among passive-like constructions, and the likely directionality of change. Section 5 concludes.

Unless otherwise noted, the data presented in this paper are drawn from primary fieldwork conducted between 2021 and 2025, comprising grammaticality judgment tasks and elicitation. Five native speakers of East Javanese were consulted, in addition to the first author’s native intuitions. All were born and raised in Surabaya and speak Indonesian as a second language; no discrepancies were observed in their judgments. The comparative data discussed in Section 4 are based on fieldwork with three monolingual Indonesian speakers born and residing in Jakarta, as well as three speakers of the North Aceh dialect of Acehnese.

2 East Javanese voice basics

Traditionally described as Indonesian-type, Javanese exhibits a three-way voice system (e.g. Suhandano 1994; Ogloblin 2005; Nurhayani 2014; Robson 2014; inter alia). The actor voice (9) is characterized by a homorganic nasal prefix (N-) and SVO word order. The preverbal field is occupied by a DP corresponding to the grammatical subject in accusative languages—i.e. the external argument in unergatives and transitives, or the internal argument in unaccusatives and passives. This DP follows any aboutness topic and precedes the aspectual auxiliary, while transitive themes remain postverbal.

    1. (9)
    1. Joko
    2. Joko
    1. wis
    2. already
    1. ng-gènḍong
    2. av-carry
    1. aḍi’-é.
    2. young.sibling-3.poss
    1. ‘Joko has carried his little brother.’ (actor voice)

The object voice (10) features a bare verb, flexible word order, and a strict person restriction on the agent. The theme DP may appear either in the pre-auxiliary field or postverbally, as indicated by the two triangle brackets in the examples. The agent is expressed as a person proclitic attached to the verb and is restricted to first or second person; third-person agents are ungrammatical in OV, as shown below.

    1. (10)
    1. <Aḍi’-é>
    2. <young.sibling-3.poss>
    1. wis
    2. already
    1. ta’/mbo’/*ḍi=Ø-gènḍong
    2. 1sg/2sg/*3=ov-carry
    1. <aḍi’-é>.
    2. <young.sibling-3.poss>
    1. ‘I/you/*s/he carried his little brother.’ (object voice)

The alleged passive construction (11) carries the prefix di-, traditionally described as a passive marker (Bintoro 1980; Badib 1980; Wedhawati & Arifin 2006; Robson 2014; Krauße 2017; inter alia; see also Van den Berg 2004 and Nurhayani 2014 for discussions of the etymology and diachronic development of di-), and cognate with the passive affix di- in Indonesian. We gloss this affix as ‘3/pass’, as it has also been analyzed as a third-person marker (e.g. Ramelan 1983; Suhandano 1994). Like OV, this construction is highly productive with dynamic bivalent verbs, displays word order flexibility, and allows the theme to surface either preverbally or postverbally, as in (11).

    1. (11)
    1. <Aḍi’-é>
    2. <young.sibling-3.poss>
    1. wis
    2. already
    1. ḍi-gènḍong
    2. 3/pass-carry
    1. <aḍi’-é>
    2. <young.sibling-3.poss>
    1. ((ambè’)
    2. (by
    1. Joko).
    2. Joko)
    1. ‘S/he/Joko carried his little brother.’ (passive voice)

However, unlike the OV construction, the di-construction permits only third-person agents, as in (12a). The prefix di- likewise cannot function as a general passive marker co-occurring with first- or second-person agent morphology, as shown in (12b). This restriction is not unique to East Javanese; it has also been noted for Javanese more broadly (e.g. Ramelan 1983; Sofwan 2001), although some dialects do not exhibit it. See Section 4 for further discussion.

    1. (12)
    1. a.
    1.   Montor-é
    2.   car-def
    1. wis
    2. perf
    1. ḍi-umbah
    2. 3/pass- wash
    1. (ambè’
    2. (by
    1. Guntur/*aku/*koen).
    2. Guntur/*1sg/2sg)
    1.   ‘The car has been washed (by Guntur/*me/*you).’
    1.  
    1. b.
    1. *Montor-é
    2.   car-def
    1. wis
    2. perf
    1. ḍi-(ta’/mbo’)-umbah
    2. 3/pass-(1sg/2sg) wash
    1. (ambè’
    2. (by
    1. aku/koen).
    2. 1sg/2sg)
    1.   (Intended: ‘The car has been washed (by *me/*you).’)

In addition to the third-person prefix di-, the agent may also be optionally expressed as a prepositional phrase (PP). When the PP is adjacent to the verb, the preposition ambe’ may be omitted, as in (13).

    1. (13)
    1. Montor-é
    2. car-def
    1. wis
    2. perf
    1. ḍi-umbah
    2. 3/pass- wash
    1. ((ambè’)
    2. ((by)
    1. Guntur).
    2. Guntur/*1sg/2sg)
    1. ‘The car has been washed (by Guntur/*me/*you).’

In short, both the passive and the OV construction employ an agent-indexing morpheme attached to the right edge of the verb stem, but they exhibit complementary person constraints, as summarized in (14). The labels “OV” and “passive” are used here in quotation marks, as we argue that the two are best analyzed as a single construction.

    1. (14)
    1. Basic traits of East Javanese voices

For clarity, we use the term ‘pivot’ throughout the remainder of the paper to refer to the phrase eligible for the preverbal position in each voice—namely, the boldfaced agent in AV (9) and the boldfaced theme in OV and the passive (10)–(11).

3 The topic vs. subject status of East Javanese pivots: A re-examination

The syntactic status of the pivot—i.e. the syntactically prominent phrase eligible to appear in specific linear positions and participate in relativization—has long been a central question in the western Austronesian literature (e.g. Schachter 1976; Shibatani 1988; Kroeger 1993; Richards 2000; Pearson 2005; Chen 2025; see also Davies 1993 for a focused discussion of Javanese). Recent work further suggests that the answer may be language-specific (e.g. Pearson 2005; Legate 2014; Patrianto & Chen 2023a; b; Lohninger & Katochoritis 2025). In this section, we present new data supporting the view that pivots in East Javanese are topics (Ā-elements) rather than grammatical subjects (derived A-elements). On this view, the pivot of the di-construction is an object topic rather than a promoted theme subject. Evidence for this claim comes from six diagnostics, which jointly show that the theme in this construction behaves as an Ā-element with topic properties, while the agent retains its role as the binder of the theme pivot.

This view is not novel for Javanese. Cole et al. (2002), for example, posit that Javanese “subjects” (i.e. pivots) are Ā-topics that cannot remain in subject position, citing their definiteness constraint and the fact that wh-subjects, like ordinary pivots, must raise to a higher position rather than remain in the grammatical subject position. See also Poedjosoedarmo (1977), Suharno (1982), Davies (1993), Ewing (2005), Sato (2015), and Vander Klok (2019) for similar observations that Javanese pivots exhibit topic-like behavior.3 By contrast, Conners (2008) and Nurhayani (2014) analyzed pivots in Tengger Javanese and Yogyakarta Javanese as grammatical subjects. Despite these competing analyses, the literature to date lacks a unified set of diagnostics for a holistic assessment of the pivot’s syntactic status and its A- versus Ā-properties—a gap that this paper aims to address.

3.1 Definiteness/specificity constraints on pivots

Empirical support for the topic analysis of pivots comes first from their strict definiteness constraint: in East Javanese, a pivot phrase must be definite or specific, and its absence results in ungrammaticality. This constraint holds regardless of the pivot’s linear position and does not extend to non-pivot phrases, irrespective of their grammatical function or thematic role, as illustrated in (15a–c). See also Poedjosoedarmo (1977), Davies (1993), Cole et al. (2002), and Vander Klok (2019) for the same observation. This systematic association between pivothood and definiteness aligns with well-established links between topicality and givenness or definiteness in the literature (Prince 1981; Lambrecht 1994; Rizzi 1997).

    1. (15)
    1. a.
    1. Wong-*(é)
    2. person-*(def)
    1. ng-guwa’
    2. av-throw.away
    1. tas-(é).
    2. bag-def
    1. ‘{The/*a} man threw {a/the} bag away.’ (AV with an agent pivot)
    1.  
    1. b.
    1. Lawuh-*(é)
    2. side.dish-*(def)
    1. ḍi-pangan
    2. 3/pass-eat
    1. (kucing
    2. (cat
    1. (iku)).
    2. (dem))
    1. ‘{The/*a} sidedish was eaten by {that/a} cat.’ (passive with a theme pivot)
    1.  
    1. c.
    1. Tas-*(é)
    2. bag-*(def)
    1. ta’/mbo’=Ø-guwa’
    2. 1sg/2sg=ov-throw
    1. (tas-*(é)).
    2. (bag-*(def))
    1. ‘I/you have thrown away {the/*a} bag.’ (OV with a theme pivot)

Crucially, this constraint applies uniquely per clause, even in constructions where more than one phrase is eligible (e.g. phrases bearing the same grammatical relation). To our knowledge, the link between pivothood and definiteness has received relatively little attention in the literature on Indonesian-type Austronesian languages (see, however, Wechsler & Arka 1998 and Erlewine et al. 2019 for brief discussion, and Davies 1993 for the same extraction constraint in Javanese). In passive ditransitives, only the object that surfaces preverbally—that is, the pivot—must satisfy the definiteness requirement. This is illustrated in (16a–b): while the preverbal object is obligatorily definite, all postverbal phrases remain indefinite. This previously underexplored pattern highlights the link between pivothood, preverbal position, and definiteness—the key indicators of topicality.

    1. (16)
    1. a.
    1. Nang
    2. prep
    1. taman,
    2. park
    1. arè’
    2. child
    1. *(iku)
    2. (dem)
    1. ḍi-kè’-i
    2. 3/pass-give-appl
    1. ḍui’-(é).
    2. money-(def)
    1. ‘In {the/a} park, s/he gave {that/*a} child {the/some} money.’ (passive)
    1.  
    1. b.
    1. Nang
    2. prep
    1. taman,
    2. park
    1. ḍui’-*(é)
    2. money-(def)
    1. ḍi-kè’-no
    2. 3/pass-give-appl
    1. arè’
    2. child
    1. (iku).
    2. (dem)
    1. ‘In {the/a} park, s/he gave {that/a} child {the/*some} money.’ (passive)

The same constraint appears in the object voice: only the object designated as the pivot must bear definite marking, while the non-pivot object may freely remain indefinite. Consider (17).

    1. (17)
    1. a.
    1. Nang
    2. prep
    1. taman,
    2. park
    1. arè’
    2. child
    1. *(iku)
    2. (dem)
    1. {ta’/mbo’}=Ø-kè’-i
    2. {1sg/2sg}=ov-give-appl
    1. ḍui’-(é).
    2. money-(def)
    1. ‘In {the/a} park, I/you gave {that/*a} child {the/some} money.’ (OV)
    1.  
    1. b.
    1. Nang
    2. prep
    1. taman,
    2. park
    1. ḍui’-*(é)
    2. money-(def)
    1. {ta’/mbo’}=Ø-kè’-no
    2. {1sg/2sg}=ov-give-appl
    1. arè’
    2. child
    1. (iku).
    2. (dem)
    1. ‘In {the/a} park, I/you gave {that/a} child {the/*some} money.’ (OV)

Notably, this constraint is also found across a range of languages with topic-indicating verbal morphology, including Tagalog (Philippine-type Austronesian), Malagasy, and the Western Nilotic languages Kurmuk and Dinka. In all such languages, each finite clause permits only a single topic-/pivot-marked phrase. See Richards (2000), Pearson (2005), Andersen (2015), and van Urk (2015) for details.

3.2 Binding parameter

Further support for analyzing East Javanese pivots as topics (Ā-elements) comes from binding diagnostics. If pivots occupied the grammatical subject position—as traditionally assumed for Indonesian-type voice systems—we would expect the theme pivot in passive constructions to function as a binder and serve as a new antecedent for anaphors (Legate 2012; van Urk 2015). Such a pattern is well attested cross-linguistically: in many languages, including English as well as the western Indonesian languages Acehnese and Indonesian, the promoted theme in a passive behaves as a new binder in the derived A-position (e.g. Arka & Manning 2008; Legate 2014). This can be tested using standard diagnostics, including quantifier–variable binding and reflexive binding. Both show that the theme pivot in all three languages behaves as a new binder and can bind a pronoun within the by-phrase, as in (18)–(20). Underlining in the translations indicates the pivot phrase.4

    1. (18)
    1. Medusai was poisoned by herselfi. (theme subject binds the agent)
    1. (19)
    1. Acehnese
    1.  
    1. a.
    1. Quantifier-variable binding: theme pivot binds the agent
    1. Tiep-tiep
    2. every
    1. aneuk
    2. child
    1. geu-lindong
    2. 3pol-protect
    1. le
    2. le
    1. mak
    2. mother
    1. droe-jih.
    2. self-3fam
    1. ‘Every childi is protected by theiri mother.’ (Legate 2014: 15)
    1.  
    1. b.
    1. Reflexive binding: theme pivot binds the agent
    1. Si
    2. art
    1. Rina
    2. Rina
    1. ji-poh
    2. 3fam-hit
    1. le droekeudroe-jih.
    2. by self-3fam
    1. ‘Rina was hit/hurt by herself.’
    1. (20)
    1. Indonesian
    1.  
    1. a.
    1. Quantifier-variable binding: theme pivot binds the agent
    1. Setiap
    2. every
    1. anak
    2. child
    1. di-sayang-i
    2. pass-love-app
    1. oleh
    2. by
    1. ibu-nya.
    2. mother-3.poss
    1. ‘Every child was loved by their mother.’
    1.  
    1. b.
    1. Reflexive binding: theme pivot binds the agent
    1. Dia/Rina
    2. 3sg/Rina
    1. di-sakit-i
    2. pass-pain-appl
    1. oleh
    2. by
    1. diri-nya
    2. body-3sg
    1. sendiri.
    2. self
    1. ‘She/Rina was hurt by herself.’

Conversely, in the East Javanese di-construction, the theme pivot cannot serve as an antecedent binding into a by-phrase, as shown in (21).5

    1. (21)
    1. East Javanese
    1.  
    1. a.
    1. Quantifier-variable binding: theme pivot unable to bind the agent
    1. *Sa’ben
    2.   every
    1. arè’
    2. child
    1. ḍi-opèn-i
    2. 3/pass-take.care.of-appl
    1. ambè’
    2. by
    1. ema’-é.
    2. mother-3.poss
    1.   (Intended: ‘Every childj is taken care of by theirj mother.’)(di-construction)
    1.  
    1. b.
    1. Reflexive binding: theme pivot unable to bind the agent
    1. *{Joko / ḍè’é}i
    2.   {Joko / 3sg}
    1. ḍii-lara-ni
    2. 3/pass-pain-appl
    1. (ambè’
    2. (by
    1. awa’-é
    2. body-3.poss
    1. ḍéwé).
    2. self)
    1.   (Intended: ‘Joko/s/he was hurt by himself/herself.’) (di-construction)

Furthermore, in the East Javanese di-construction, the theme pivot may instead surface as a reflexive bound by a third-person agent, as in (22), a pattern not attested in Indonesian or Acehnese.

    1. (22)
    1. East Javanese
    1.  
    1. a.
    1. Quantifier-variable binding: theme pivot bound by the agent
    1. Omah-omah-éi/j
    2. house-red-3.poss
    1. ḍi-cet
    2. 3/pass-paint
    1. ambè’
    2. by
    1. sa’ben
    2. every
    1. wongi.
    2. person
    1. ‘Theiri/j houses were painted by every personi.’(di-construction)
    1.  
    1. b.
    1. Reflexive binding: theme pivot bound by the agent
    1. [Awa’-é
    2. [body-3.poss
    1. ḍéwé]i
    2. self]
    1. ḍii-lara-ni
    2. 3/pass-pain/sick-appl
    1. (ambè’
    2. (by
    1. Joko).
    2. Joko)
    1. Lit: ‘Herself/himself was hurt by Joko/him/her.’ (di-construction)

This pattern indicates that the theme pivot is not the grammatical subject of the di-construction—contra recent analyses of theme arguments in Acehnese and Indonesian passives (19)–(20)—but instead behaves as an Ā-element, exhibiting reconstruction effects and being interpreted in its θ-position. This conclusion is consistent with the observation in Section 3.1 that the theme pivot patterns as a topic.

The same pattern is observed in the East Javanese OV construction: the theme pivot cannot function as a binder (23a) but may instead surface as a reflexive (23b), exactly as in the di-construction (21)–(22).

    1. (23)
    1. a.
    1. Reflexive binding: theme pivot unable to bind the agent
    1. *Aku/koen
    2. 1sg/2sg
    1. ta’/mbo’=Ø-lara-ni.
    2. 1sg/2sg=ov-hurt-appl
    1. (Intended: ‘I/you were hurt by me/you.’) (object voice)
    1.  
    1. b.
    1. Reflexive binding: agent binds theme pivot
    1. Awa’-ku/-mu
    2. body-1sg/2sg
    1. ḍéwé
    2. self
    1. ta’/mbo’=Ø-lara-ni.
    2. 1sg/2sg=ov-hurt-appl
    1. ‘I/you hurt myself/yourself.’ (object voice)

The East Javanese AV construction shows the same pattern: the agent asymmetrically binds the theme, as in (24a–b).

    1. (24)
    1. a.
    1. Reflexive binding: agent pivot binds the theme
    1.   Aku
    2.   1sg
    1. nulung
    2. av.help
    1. awa’-ku ḍéwé.
    2. body-1sg self
    1.   ‘I helped myself.’(actor voice)
    1.  
    1. b.
    1. Reflexive binding: theme unable to bind agent pivot
    1. *Awa’-ku
    2.   body-1sg
    1. ḍéwé
    2. self
    1. nulung
    2. av.help
    1. aku.
    2. 1sg
    1.   (Intended: ‘Myself helped me.’(actor voice)

This observation indicates that voice alternation in East Javanese does not affect binding relations across the three voice constructions. This invariant pattern supports our analysis that pivots function as Ā-elements (topics), whose promotion is not expected to introduce new binding antecedents.6 It also aligns with the fact that pivots are subject to the definiteness constraint commonly associated with topics, as discussed earlier in Section 2.2.

3.3 Prepositional phrases’ eligibility to constitute the pivot

The third diagnostic for assessing the syntactic status of pivots concerns their compatibility with prepositional phrases (PPs). If pivots are probed by [uD] and occupy a derived A-position as grammatical subjects—as proposed for Indonesian and Acehnese (e.g. Aldridge 2008; Cole et al. 2008; Legate 2012)—PPs should not be able to serve as pivots.

If, however, pivots are probed by an Ā-probe (i.e. [utop], as we propose) and thus occupy an Ā-position, a PP bearing a topic feature should also be eligible for pivot status (Miyagawa 2010; van Urk 2015). Recall that under our Ā-based analysis of East Javanese voice, actor voice involves subject topicalization, whereas the di-construction and object voice involve object topicalization (8). The resulting prediction is that PPs should not receive pivot status in AV (subject topicalization), but should be eligible to do so in the di-construction and object voice, which involve object topicalization.

This prediction is borne out. In the East Javanese di-construction, a definite PP can surface preverbally and satisfy the definiteness constraint associated with pivothood (Section 3.1). Crucially, when such a PP occupies the preverbal position, any theme DP must remain postverbal and need not be definite-marked, as illustrated in (25). This pattern indicates that the PP—not the postverbal theme—is the true pivot, given that pivothood in East Javanese is strictly conditioned by definiteness and specificity.

    1. (25)
    1. [nang
    2. [prep
    1. kebun *(iku)]
    2. garden (that)]
    1. até
    2. fut
    1. ḍi-tandur
    2. 3/pass-plant
    1. pirang-pirang
    2. several-red
    1. kembang.
    2. flower
    1. ‘In *(that) garden she/he is going to plant several flowers.’ (di-construction)

There is independent evidence that the clause-initial PP is a true pivot rather than an aboutness topic or a preposed adjunct. In East Javanese, each clause permits at most one aboutness topic, which must appear to the left of the pivot (Davies 1993).7 Unlike pivots, aboutness topics are not subject to the definiteness constraint, as shown in (26). The abbreviation atop denotes ‘aboutness topic’.

    1. (26)
    1. [Nang
    2. [prep
    1. kebun-(é)]ATOP
    2. garden-(def)]
    1. [wong
    2. [man
    1. *(iku)]pivot
    2. dem]
    1. nandur
    2. av.plant
    1. pirang-pirang
    2. several-red
    1. kembang.
    2. flower
    1. ‘In {the/any} garden, {the/*a} man planted several flowers.’ (actor voice)

In the di-construction, a PP may intervene between an aboutness topic and an aspectual auxiliary (e.g. ate), as in (27). Crucially, like DP pivots, such PPs must be definiteness-marked, unlike aboutness topics (26). This rules out analyzing the obligatorily definite PP (‘in the garden’) as a second aboutness topic, thereby reinforcing the view that the preverbal PP is a genuine pivot rather than an aboutness topic or adjunct, since both aboutness topics and adjuncts are exempt from the definiteness constraint.

    1. (27)
    1. [Pirang-pirang
    2. [several-red
    1. kembang]ATOP
    2. flower]
    1. [nang
    2. [prep
    1. kebun *(iku)/(*nḍi
    2. garden (that)/(which
    1. aé)]
    2. ae)]
    1. até
    2. fut
    1. ḍi-tandur.
    2. 3/pass-plant
    1. ‘Several flowers, in {that/*any} garden, she/he is going to plant (them).’

Further support for analyzing such PPs as pivots comes from di-constructions that remain grammatical when the PP is the sole definite phrase in the clause, as in (28). This pattern confirms that the PP is the true pivot, consistent with the strict definiteness constraint on pivots in East Javanese.

    1. (28)
    1. [Nang
    2. [prep
    1. pinggir
    2. side
    1. dalan]ATOP
    2. road]
    1. [ambè’
    2. [prep
    1. lading
    2. knife
    1. iki]pivot
    2. dem]
    1. iso
    2. mod
    1. meneng-meneng
    2. silent-red
    1. ḍi-keṭo’-i
    2. 3/pass-cut-appl
    1. wong
    2. person
    1. kabel
    2. cable
    1. listrik.
    2. electricity
    1. ‘Near a roadside, one can silently cut an electric cable with this knife.’

Furthermore, PP pivots exhibit reconstruction effects, on a par with DP pivots, as shown in (29).

    1. (29)
    1. [Tèmbo’]ATOP
    2. [wall]
    1. [ambè’
    2. [prep
    1. foto-né
    2. photo-3.poss
    1. awa’-′e
    2. body-3.poss
    1. ḍéwé]pivot
    2. self]
    1. ḍi-tèmpèl-i
    2. 3/pass-stuck-appl
    1. Priska.
    2. Priska
    1. ‘A wall, Priskai covered (it) with a photo of herselfi/k.’

The OV construction likewise permits PP pivots, as illustrated in (30), where a PP pivot surfaces between an indefinite aboutness topic and an aspectual auxiliary and serves as the sole definite phrase in the clause.

    1. (30)
    1. [Pirang-pirang
    2. [several-red
    1. kembang]ATOP
    2. flower]
    1. [nang
    2. [prep
    1. kebun *(iku)/(*nḍi
    2. garden (that)/(which
    1. aé)]
    2. ae)]
    1. até
    2. fut
    1. {ta’/mbo’}=Ø-tandur.
    2. {ov:1sg/2sg}-ov-plant
    1. ‘Several flowers, in {that/*any} garden, I/you am/are going to plant (them).’

Conversely, the AV construction does not allow a PP to surface in the pivot position, as shown in (31). This follows from our analysis that AV involves subject topicalization, which is expected to exclude PP pivots, as PPs cannot satisfy [uD] and therefore cannot be promoted to subject.

    1. (31)
    1. *[Joko]ATOP
    2.   [Joko]
    1. [nang
    2. [prep
    1. omah-é]intended pivot
    2. house-def]
    1. até
    2. fut
    1. moco
    2. av.read
    1. buku.
    2. book
    1.   (Intended: ‘As for Joko, in the house (he) will be reading a book.’) (actor voice)

The availability of PP pivots in both the di- and OV constructions therefore supports the topic analysis of pivots and challenges traditional accounts of Javanese voice that treat pivots as grammatical subjects probed by [uD]. Notably, such PP pivots may embed a DP bearing a range of thematic roles, including instrument (32a), locative (32b), reason (32c), beneficiary (32d), and comitative (32e). This pattern closely parallels Locative and Circumstantial Voice constructions in Philippine-type languages, which likewise permit peripheral, non-selected phrases to be promoted to pivot status, including instruments, locatives, reasons, and other adjuncts also attested in the East Javanese OV construction. See Chen & McDonnell (2019) for an overview of such voice systems and Rackowski (2002) for a detailed discussion of Tagalog.

    1. (32)
    1. PP pivots in the East Javanese passive/object voice constructions
    1.  
    1. a.
    1. Ambè’
    2. with
    1. hapé-*(né)
    2. cellphone-3.poss
    1. {ḍi-/ta’/mbo’}=jupu’
    2. {3/pass-/ov:1sg/2sg}=take
    1. sembarang
    2. any
    1. gambar.
    2. picture
    1. ‘I/you/s/he took a picture with her/his/*a cellphone.’
    1.  
    1. b.
    1. Nang
    2. to
    1. omah-*(é)
    2. house-3.poss
    1. {ḍi-/ta’/mbo’}=kirim
    2. {3/pass-/ov:1sg/2sg}=send
    1. surat
    2. letter
    1. opo
    2. what
    1. aé.
    2. ae
    1. ‘I/you/s/he sent any letter to her/his/*a house.’
    1.  
    1. c.
    1. Gara-gara
    2. because
    1. utang-*(é)
    2. debt-3.poss
    1. {ḍi-/ta’/mbo’}=jalu’-i
    2. {3/pass-/ov:1sg/2sg}=ask.for-appl
    1. ḍui’
    2. money
    1. sopo
    2. who
    1. aé.
    2. ae
    1. ‘I/you/s/he asked any person for money because of her/his/*some debt.’
    1.  
    1. d.
    1. Kanggo
    2. for
    1. {Joko/arè’-*(é)}
    2. {Joko/child-3.poss}
    1. {ḍi-/ta’/mbo’}=buka’-no
    2. {3/pass-/ov:1sg/2sg}=open-appl
    1. lawang
    2. door
    1. nḍi
    2. which
    1. aé
    2. ae
    1. ‘I/you/s/he opened any door for {Joko/the/*a boy}.’
    1.  
    1. e.
    1. Ambè’
    2. with
    1. {Maria/arè’-*(é)}
    2. {Mary/child-3.poss}
    1. {ḍi-/ta’/mbo’}=resi’-i
    2. {3/pass-/ov:1sg/2sg}=clean-appl
    1. omah
    2. house
    1. nḍi
    2. which
    1. aé.
    2. ae
    1. ‘I/you/s/he cleaned any house with {Mary/the/*a boy}.’

3.4 Flexibility in pivot selection

Additional support for the topic analysis of pivots comes from ditransitive constructions. In East Javanese, ditransitives exhibit flexible pivot selection when more than one object or PP is present. In both the di-construction and object voice, either an adjunct PP or one of the two objects may surface between the aboutness topic and the aspectual auxiliary and serve as the pivot. As expected, such PPs must be definite and specific (33a–c), exhibiting a defining property of pivot phrases.

    1. (33)
    1. a.
    1. [Nang
    2. [prep
    1. warung
    2. restaurant
    1. (iku)]ATOP
    2. dem]
    1. wong
    2. person
    1. wèdo’
    2. female
    1. *(iku)
    2. dem
    1. {ḍi-/ta’/mbo’}=kè’-i
    2. {3/pass-/ov:1sg/2sg}=give-appl
    1. (wong
    2. (person
    1. wèdo’
    2. female
    1. *(iku))
    2. dem)
    1. [ḍui’]
    2. [money]
    1. [nang
    2. [prep
    1. mèjo-(é)].
    2. table-def]
    1. ‘In {a/the} restaurant, I/you/s/he gave {the/*a} woman {some} money on {her/a} table.’
    1.  
    1. b.
    1. [Nang
    2. [prep
    1. warung]ATOP
    2. restaurant]
    1. nang
    2. prep
    1. mèjo
    2. table
    1. *(iku)
    2. dem
    1. {ḍi-/ta’/mbo’}=kè’-i
    2. {3/pass-/ov:1sg/2sg}=give-appl
    1. ḍui’
    2. money
    1. (nang
    2. (prep
    1. mèjo
    2. table
    1. *(iku))
    2. dem)
    1. [pirang-pirang
    2. several-red
    1. wong
    2. person
    1. wèdo’].
    2. female
    1. ‘In{a/the} restaurant, I/you/s/he gave some women {some} money on {that/*a} table.’
    1.  
    1. c.
    1. [Nang
    2. prep
    1. mèjo]ATOP
    2. table
    1. nang
    2. (prep
    1. warung
    2. restaurant
    1. *(iku)
    2. dem)
    1. {ḍi-/ta’/mbo’}=kè’-i
    2. {3/pass-/ov:1sg/2sg}=give-appl
    1. ḍui’
    2. money
    1. wong
    2. person
    1. wèdo’
    2. female
    1. nḍi
    2. which
    1. aé
    2. ae
    1. (nang
    2. (prep
    1. warung
    2. restaurant
    1. *(iku)).
    2. dem)
    1. ‘On {her/a} table, I/you/s/he gave {the/a} woman {some} money in {the/*a} restaurant.’

This flexibility in pivot selection reinforces the view that promotion to pivothood in East Javanese is not constrained by the locality requirements of [uD] and does not involve promotion to subject, which must respect locality. This nonlocality of pivot selection therefore further undermines traditional A-approaches to Javanese voice and supports a topicalization analysis, under which both non-subject DPs and PPs are eligible pivots.

4 Illusory “passive” as nonsubject topicalization

We now reconsider the structure of the di-construction (34), shown in Section 3 to involve Ā-movement of a non-subject phrase.

    1. (34)
    1. Tahu-né
    2. tofu-def
    1. wis
    2. already
    1. ḍi-pangan
    2. 3/pass-eat
    1. (ambè’
    2. (by
    1. konco-ku).
    2. friend-1sg.poss)
    1. ‘S/he/my friend ate the tofu.’

Recall that this construction shares three core properties with object voice: (i) both display a verbal affix indexing the person and number of the agent or initiator, (ii) both permit PP pivots, and (iii) in both, the pivot may appear either preverbally or postverbally. This shared configuration is schematized in (35).

    1. (35)
    1. (pivot) (auxiliary) prefix{1/2/3}-V (pivot) nonpivot phrases (pivot)

We propose that both constructions involve a non-subject topic—either a DP or a PP—that Ā-moves directly from its θ-position. In both cases, a non-subject phrase bears a [top] feature and enters into an Agree relation with [utop], triggering Ā-movement to a left-peripheral topic position, as illustrated in (36). Following previous work on Javanese (Cole et al. 2002; Vander Klok 2019), we assume this position to be above FocusP and above the derived A-position in East Javanese triggered by Agree with [uD].

On this view, the di-construction is an object topicalization construction with a third-person agent, while what has traditionally been analyzed as object voice involves a first- or second-person agent. This analysis builds on earlier insights from Uhlenbeck (1975) and Poedjosoedarmo (1986), both of whom note that Javanese is a topic-oriented language in which discourse topic largely determines sentence structure.8 See also Asikin-Garmager (2017) and Khairunnisa (2022) for similar topic analyses of pivots in Sasak, Nagaya (2012) for Lamaholot, and Sirima (2025) for Mori Atas.

    1. (36)
    1. Javanese OV/di-construction as involving nonsubject topicalization

The theoretical assumptions behind this analysis are as follows. We assume that all finite clauses in East Javanese bear an uninterpretable topic feature on an Ā-head, which probes the closest phrase with a matching feature. Given Relativized Minimality (Rizzi 1990), this Agree relation skips any intervening DPs and PPs lacking a topic feature. A topic phrase therefore need not be the highest phrase in the clause to agree with [utop].

    1. (37)
    1. Relativized Minimality
    2. A syntactic relation R must involve the closest XP capable of entering into R.

Where the highest DP in a clause also bears a topic feature, it is first probed by [uD] on T and raises to the grammatical subject position, before undergoing Ā-movement to [Spec TopP] following its Agree relation with [utop]. Following earlier work on Malagasy (Pearson 2005), we treat the AV affix N- as extraction morphology indexing the nominative case status of the subject topic.9

Where the topic of a clause is a non-subject phrase (either an object or a PP), that phrase enters into an Agree relation with [utop] and Ā-moves to [Spec TopP], while the highest phrase in the clause agrees with [uD] and A-moves to the grammatical subject position, as illustrated in (36). We propose that no overt extraction morphology is associated with non-subject topicalization, since the phrase undergoing Ā-movement does not bear nominative case.

The fact that the di-construction patterns with object voice in quantifier float—disallowing pre-auxiliary stranding, in contrast to actor voice—follows naturally from this analysis. This is shown in (38), which demonstrates a quantifier float asymmetry between AV (38a) and the two constructions under discussion (38b–c), paralleling the pattern described in Section 1.10

    1. (38)
    1. a.
    1.   [___i
    2.   [___
    1. Konco-ku]
    2. friend-1sg]
    1. telui
    2. three
    1. até
    2. fut
    1. ng-gawé
    2. av-make
    1. layangan.
    2. kite
    1.   ‘My three friends will make kites.’ (actor voice)
    1.  
    1. b.
    1. *[___i
    2.   [___
    1. Montor-é]
    2. car-def]
    1. telui
    2. three
    1. até
    2. fut
    1. ta’=Ø-dandan-i.
    2. 1sg=ov-fix-appl
    1.   (intended: ‘I will fix three of the cars.’) (object voice)
    1.  
    1. c.
    1. *[___i
    2.   [___
    1. Montor-é]
    2. car-def]
    1. telui
    2. three
    1. até
    2. fut
    1. ḍi-dandan-i
    2. 3/pass-fix-appl
    1. ((ambè’)
    2. by
    1. konco-ku).
    2. friend-1sg.poss
    1.   (intended: ‘Three cars will be fixed (by s/he/my friend).’) (passive)

A further asymmetry in East Javanese quantifier float supports our analysis: the OV construction and the di-construction—but not AV—permit quantifier stranding in postverbal position, as in (39a–c). This follows from our analysis: in AV, the external argument is base-generated in preverbal position and undergoes A-movement, leaving no postverbal position available for stranding.11

    1. (39)
    1. a.
    1.   [___i
    2.   [___
    1. Tahu-né]
    2. tofu-def
    1. wis
    2. perf
    1. ḍi-pangan
    2. 3/pass-eat
    1. rolasi
    2. twelve
    1. ((ambè’)
    2. ((by)
    1. konco-ku).
    2. friend-1sg)
    1.   ‘Twelve of the tofu were eaten (by my friend). (passive voice)
    1.  
    1. b.
    1.   [___i
    2.   [___
    1. Tahu-né]
    2. tofu-def]
    1. wis
    2. perf
    1. ta’=Ø-pangan
    2. 1sg=ov-eat
    1. rolasi.
    2. twelve
    1.   ‘I have eaten twelve of the tofu.’(object voice)
    1.  
    1. c.
    1. *[___i
    2.   [___
    1. Konco-né]
    2. friend-3sg]
    1. até
    2. fut
    1. m-angan
    2. av-eat
    1. rolasi
    2. twelve
    1. tahu.
    2. tofu
    1.   (Intended: ‘Twelve of his/her friends will eat tofu.’)(actor voice)

There is independent evidence that the QF patterns discussed here represent genuine cases of quantifier stranding derived from Ā-movement (e.g. Sportiche 1988; Miyagawa 1989; Fitzpatrick 2006). First, as shown in the examples in Section 1 and above, QF in Javanese yields a non-exhaustive (i.e. partitive) reading, a hallmark of Ā-movement-derived quantifier stranding (Fitzpatrick 2006). Second, in postverbal positions across the three voices, the QF pattern follows the expected distribution: numeral quantifier stranding from the pivot is disallowed in AV but consistently permitted in OV and di-constructions. This pattern supports the presence of Ā-movement from the θ-position of the pivot in these constructions.

An immediate implication of this analysis is that the complementary person constraint attributed to the so-called passive and object voice is best understood as a consequence of terminological convention rather than a genuine syntactic distinction. The first- and second-person affixes traditionally described as ergative proclitics in OV, and the affix labeled as passive in the di-construction, are in fact subject clitics (or subject agreement) attached to the verb, as in (40).

    1. (40)
    1. Ā-topic (auxiliary) subject-indexing affix{1/2/3} -Verb . . .

Under (40), we assume that the alleged passive in East Javanese is a true transitive and involves no valency-decreasing operations. The fact that the agent/initiator in this construction consistently binds the theme, but not vice versa (section 3.2), follows from this analysis. In this view, di- is not a valency-decreasing affix but a third-person subject clitic encoding the nominative DP in [Spec TP]. This analysis is supported not only by existing descriptions of di- as a third-person affix, but also by the observation that (i) the di- construction is incompatible with first- and second-person agents, as in (41a), and (ii) in the absence of an overt DP/PP, all five native speakers we consulted interpret the construction as having a third-person subject, as in (41b).

    1. (41)
    1. a.
    1. *Klambi-né
    2.   shirt-def
    1. lagi
    2. prog
    1. ḍi-pépé
    2. 3/pass-hang.to.dry
    1. (ambè’)
    2. by
    1. aku/koen.
    2. 1sg/2sg
    1.   (Intended: ‘The shirt is being hung to dry by me/you.’)
    1.  
    1. b.
    1.   Tahu-ne
    2.   tofu-def
    1. wis
    2. perf
    1. ḍi-pangan
    2. 3/pass-eat
    1. (ambè’
    2. by
    1. konco-ku).
    2. friend-1sg
    1.   ‘S/he/my friend ate the tofu.’

We now turn to our proposal that East Javanese possesses an ordinary grammatical subject position, which is a derived A-position probed by [uD] and attracts the highest caseless DP along with nominative case assignment, as schematized earlier in (42) and repeated below. We assume that the prefixal person morpheme in the East Javanese OV construction (which comprises the di-construction) indexes the DP that A-moves to this position.

    1. (42)
    1. Proposed structure of the East Javanese OV/di-construction

This proposal differs from traditional analyses of Indonesian-type object voice, which analyze the agent proclitic as an ergative argument that remains in its base-generated position in [Spec VoiceP] (Aldridge 2008; Cole et al. 2008; Legate 2014). We instead argue for a subject analysis, motivated by both empirical and theoretical considerations. First, since promotion to pivothood in East Javanese is a purely Ā-operation, the grammar must make available a derived A-position distinct from the pivot position. Second, the quantifier-floating facts discussed above indicate the availability of such a derived A-position in the actor voice; given that the AV and OV differ only in whether topicalization targets a subject or a nonsubject, there is no principled reason to deny the same position in the OV construction. Finally, binding diagnostics show that the non-pivot agent in OV systematically functions as a binder of the theme, indicating that it occupies a grammatical subject position. Together, these considerations necessitate the postulation of a grammatical subject position structurally lower than the pivot position.12

We now turn to a related question concerning the status of the by-phrase in the di-construction. Following recent analyses of comparable by-phrases in Balinese (Nomoto 2018), we propose that the agent by-phrase is best analyzed as an instance of clitic doubling or agreement doubling—a pattern disallowed for first- and second-person agents/initiators to avoid redundancy. This restriction reflects a broader person-based constraint in Western Austronesian languages, whereby transitive agents encoded through person-indexing on the verb cannot also be realised as full DPs when they are first or second person.13 Consider, for example, the same constraint in the Western Austronesian language Puyuma (43), as well as a parallel restriction in Standard Indonesian (44), where the di-construction disallows first- or second-person by-phrases.

    1. (43)
    1. Puyuma: clitic doubling banned with 1st- and 2nd-person agent
    1.  
    1. a.
    1. {ku / nu}=trakaw-aw
    2. {1SG.GEN / 2sg.gen}=steal-pv
    1. na
    2. def.pivot
    1. aputr
    2. flower
    1. *(kanku/kanu).
    2. *(1sg/2sg.obl)
    1. ‘I/you stole the flowers.’ (clitic doubling banned for 1st/2nd-person agent)
    1.  
    1. b.
    1. tu=trakaw-aw
    2. 3.gen=steal-pv
    1. na
    2. def.pivot
    1. aputr
    2. flower
    1. (kan sawagu).
    2. (SG.OBL Sawagu)
    1. ‘S/he/(Sawagu) stole the flowers.’(clitic doubling with 3rd-person agent)
    1. (44)
    1. Standard Indonesian: Restriction on 1st- and 2nd-person agents in by-phrases14
    1.  
    1. a.
    1. *Buku
    2.   book
    1. itu
    2. det
    1. di-baca
    2. pass-read
    1. oleh
    2. by
    1. saya/kamu.
    2. 1sg/2sg
    1.   (Intended: ‘The book was read by me/you.’)
    1.  
    1. b.
    1.   Buku
    2.   book
    1. itu
    2. det
    1. di-baca
    2. pass-read
    1. oleh
    2. by
    1. Ali.
    2. Ali
    1.   ‘The book was read by Ali.’

Old Javanese texts suggest that the by-phrase in Javanese was historically a full NP, with the synchronic agent-introducing preposition ambè’ developing from the case marker ni (Poedjosoedarmo 2002). This marker is a regular reflex of the Proto-Austronesian genitive case *ni (Blust 2015) and has been shown to exhibit the distribution typically associated with structural nominative case in conservative Austronesian languages that retain it (Chen 2025; see also Rackowski 2002; Rackowski & Richards 2005 for earlier nominative analyses). This supports the present proposal that the East Javanese pronominal affix indexes the nominative argument, which was originally realized as a full DP and later reanalyzed as an optional by-phrase through the development of the preposition.15

Notably, similar diachronic developments—whereby a case-marked core NP is reanalyzed as an oblique—are attested in other Austronesian languages and are reflected in synchronic dialectal variation. Consider the data from two dialects of Puyuma below. In the more conservative Ulivelivek dialect (45a), non-pivot agents carry the genitive case marker ni; in the more innovative Nanwang dialect (45b), such agents instead bear the oblique marker kan/kana, which typically marks adjuncts, as illustrated in (45c) (Teng 2009). We suggest that the East Javanese di-construction may have undergone a similar development.

    1. (45)
    1. Puyuma: reanalysis of nonpivot agent marking into an oblique marker
    1.  
    1. a.
    1. tawi=pinatray-aw
    2. 3.gen=kill-pv
    1. i
    2. sg.pivot
    1. tava
    2. Tava
    1. [ni
    2. [sg.gen
    1. takiyu]i
    2. Takiyu]
    1. ‘Takiyu killed Tava.’ (Teng 2009: 824; glosses ours)(Ulivelivek Puyuma)
    1.  
    1. b.
    1. tui=padrek-aw
    2. 3.gen=carry.on.back-pv
    1. i
    2. sg.pivot
    1. atrung
    2. Atrung
    1. [kan
    2. [sg.obl
    1. senten
    2. Senten
    1. /
    2. /
    1. kana
    2. def.obl
    1. walak]i.
    2. child]
    1. ‘{Senten/the child} carried Atrung (on her/their back).’ (Nanwang Puyuma)
    1.  
    1. c.
    1. taw=pinatray-aw
    2. 3.gen=kill-pv
    1. i
    2. sg.pivot
    1. tava
    2. Tava
    1. (kana
    2. (def.obl
    1. tazaw)
    2. knife)
    1. ‘He killed Tava (with the knife).’ (Teng 2009: 824) (Ulivelivek Puyuma)

To conclude, the East Javanese data motivate an analysis in which the language exhibits a two-way voice system encoding topicalization, with no independent passive construction, as summarized in (46).

    1. (46)
    1. East Javanese voice as a two-way Ā-oriented topic-indexing system
    1.  
    1. a.
    1. Actor voice: subject topic construction, marked by the homorganic nasal prefix N-, with no person-indexing proclitic (i.e. subject-indexing verbal affix)
    1.  
    1. b.
    1. Object voice: object topic construction, unmarked, with an obligatory subject-indexing affix

We suggest that the absence of a person-indexing proclitic (i.e. subject-indexing verbal affix) in the East Javanese AV construction is the result of an anti-agreement effect (Ouhalla 1993; Baier 2018). This effect is expected, given that the grammatical subject in a subject-topic construction undergoes further Ā-extraction to an Ā-position—an environment in which such effects are predicted to arise. We do not pursue this asymmetry further, as it lies beyond the scope of the present paper.

5 Implications: “Indonesian-type passives” as a cline

The present observations from East Javanese raise two further questions. First, can the passive constructions reported in neighboring Austronesian languages—such as Indonesian and Acehnese—also be accounted for under the current analysis? Second, do other Javanese dialects exhibit a similar Ā-oriented pseudo-passive construction that involves no A-movement of the theme to the grammatical subject position?

New comparative data suggest that the answer to the first question is negative. Our primary fieldwork confirms that pivots in Indonesian and Acehnese behave fundamentally differently from those in Javanese—not only in the binding patterns reported in the literature (Section 3.3), but also in their exemption from the definiteness constraint and their incompatibility with PP pivots, consistent with existing analyses of both languages. Consider the examples in (47)–(48).16 Together with the binding patterns indicating that the pivot functions as the subject, these findings reinforce the view that passives in Indonesian and Acehnese involve genuine A-movement to the grammatical subject position (Cole et al. 2008; Legate 2014).

The East Javanese case thus highlights surface-level similarities between object topicalization and passivization in languages lacking morphological case, while also revealing underappreciated variation within so-called Indonesian-type three-way voice systems. These systems are not structurally uniform but instead form a cline of voice systems in flux, in line with recent work highlighting syntactic variation among Western Austronesian languages roughly classified as Indonesian-type (Hemmings 2015; 2021; Himmelmann 2020; McDonnell & Chen 2022; Patrianto & Chen 2023a; inter alia).

    1. (47)
    1. a.
    1. Se-buah
    2. one-clf
    1. buku
    2. book
    1. di-temu-kan
    2. pass-find-appl
    1. di
    2. prep
    1. sampah.
    2. garbage
    1. ‘A book was found in a rubbish bin.’(Colloquial Indonesian)17
    1.  
    1. b.
    1. (Sa-boh)
    2.   one-clf
    1. komputer
    2. computer
    1. geu-pakek
    2. 3sgpol-use
    1. le
    2. by
    1. guree
    2. teacher
    1. nyan.
    2. dem
    1. ‘A computer was used by the teacher.’ (Acehnese)
    1. (48)
    1. a.
    1. *Untuk
    2.   prep
    1. Rina
    2. Rina
    1. kau-beli-kan
    2. 2sg-buy-appl
    1. sebuah
    2. a
    1. sepeda
    2. bicycle
    1. motor.
    2. motor
    1.   (Intended: ‘You bought a motorcycle for Rina.’)(Colloquial Indonesian)
    1.  
    1. b.
    1. *Ngen
    2.   with
    1. kamera
    2. camera
    1. lon
    2. 1sg
    1. cok
    2. take
    1. padumboh
    2. several/some
    1. poto.
    2. photo
    1.   ‘I took some photos with a camera.’(Acehnese)

Turning to the second question, a comparative examination of other Javanese dialects suggests that the East Javanese pattern described here may be conservative. Data from Tengger Javanese indicate that the di-construction pivot is not subject to a definiteness constraint. Consider (49).

    1. (49)
    1. Kucing
    2. cat
    1. ḍi-cokot
    2. pass-bite
    1. kirik.
    2. dog
    1. ‘A cat was bitten by a/the dog.’ (Conners 2008: 1654, 1565)(Tengger Javanese)

In Yogyakarta Javanese, the theme pivot does not show reconstruction effects and cannot surface as a reflexive bound by the agent, as shown in (50). This contrasts with the binding pattern observed in East Javanese (51), and suggests that the dialect may possess a pattern parallel to Indonesian.

    1. (50)
    1. *[Awak-e
    2.   body-poss
    1. ḍewe]i
    2. self
    1. ora
    2. no
    1. tau
    2. ever
    1. ḍi-pikir-ake
    2. pass-think-appl
    1. dening
    2. by
    1. Anii.
    2. Ani
    1.   (Intended: ‘Ani never thinks about herself.’) (Nurhayani 2014: & 124)
    2.   (Yogyakarta Javanese)
    1. (51)
    1. [Awa’-é
    2. [body-3.poss
    1. ḍéwé]i
    2. self]
    1. ḍii-jiwit-i
    2. 3/pass-pinch-appl
    1. (ambè’
    2. (by
    1. Joko).
    2. Joko)
    1. ‘Herself/himself was being pinched by Joko/him/her.’ (East Javanese)

A further locus of variation is observed in Surakarta Javanese, where the di-construction is reported to exhibit no restriction on first- or second-person agents, as in (52), again aligning with the Indonesian pattern. Pesisiran Javanese likewise shows a relaxed person constraint: recent data indicate that the prefix di- may cross-reference a first-person by-phrase, as in (53).

    1. (52)
    1. Surat-é
    2. letter-def
    1. ḍi-tulis
    2. di-write
    1. ḍening
    2. by
    1. {aku/kowe/aku
    2. {1sg/2sg/1st
    1. sakloron}.
    2. two}
    1. ‘The letter was written by me/you/us two.’ (Cui & Rabinovitch 2024)
    2. (Surakarta Javanese)
    1. (53)
    1. Sego
    2. rice
    1. kuwi
    2. the
    1. wis
    2. pst
    1. ḍi-masak
    2. pass-cook
    1. karo
    2. by
    1. aku.
    2. 1sg
    1. ‘The rice has been cooked by me.’ (Vander Klok 2008: 2) (Pesisiran Javanese)

To our knowledge, existing descriptions of these dialects do not address all the diagnostics discussed here, nor the quantifier float pattern; thus, whether the di-construction in these varieties is a true passive remains uncertain and warrants further investigation.

Nevertheless, the coexistence of A- and Ā-oriented di-constructions in Javanese suggests a possible derivational relationship between the two, and there is strong evidence that the East Javanese voice system is conservative. There is broad consensus in the Austronesian literature that Indonesian-type voice systems are more innovative than Philippine-type systems (Himmelmann 2002; McDonnell & Chen 2022), the latter of which can be traced back to Proto-Austronesian or a stage immediately following its split (Ross 2009). The binding pattern observed in East Javanese passive and OV constructions is consistent with patient voice constructions in Philippine-type languages such as Malagasy, Tagalog, and several Formosan languages. These languages feature a non-demoted agent and a theme pivot that exhibit the same binding behavior as the East Javanese passive/OV constructions (Tagalog: Rackowski 2002; Malagasy: Pearson 2005; Formosan languages: Chen 2017). The Tagalog examples in (54) illustrate this pattern: the pivot does not serve as a new antecedent for anaphors and is interpreted in its θ-position.

    1. (54)
    1. Tagalog
    1.  
    1. a.
    1.   hindi
    2.   neg
    1. p<in>igil
    2. <pv.pfv> control
    1. ni
    2. pn.nom
    1. Rica
    2. Rica
    1. ang
    2. pivot
    1. sarili
    2. self
    1. niya
    2. 3sg.poss
    1. (na
    2. (lk
    1. k<um>ain).
    2. eat<av>)
    1.   ‘Rica cannot stop himself (from eating).’ (patient voice)
    1.  
    1. b.
    1. *sa-sampal-in
    2.   cont-slap-pv
    1. ng
    2. cn.nom
    1. kanyang
    2. 3sg
    1. sarili
    2. refl
    1. si
    2. pn.pivot
    1. Rica.
    2. Rica
    1.   (Intended: Himself will slap Rica.’)(patient voice)

Second, the third-person constraint observed in the East Javanese di-construction—complementary to the person constraint in object voice, which is indexed by first- and second-person prefixes—suggests an economy-driven pattern that is likely to reflect an earlier system. Similar analyses have been proposed for Balinese and Malay/Indonesian, where the passive marker is argued to have originated as a third-person agent affix that was subsequently reinterpreted as a general passive marker (Van den Berg 2004; Arka 2008; Artawa 2013; Nomoto 2018; see also references therein).

Taken together, these observations suggest that many Javanese varieties may be shifting toward an Indonesian-style A-oriented voice system with a true passive, while East Javanese represents a more conservative variety that preserves Philippine-type syntax. East Javanese thus contributes to the growing literature on syntactic variation among Austronesian languages traditionally classified as Indonesian-type (see Ross 2002; Kaufman 2009; McDonnell & Chen 2022 for overviews showing that western Indonesian Austronesian languages do not form a coherent group, but instead exhibit recurrent patterns likely resulting from parallel innovations and language contact), and underscores that these voice systems should not be treated as a homogeneous group.

6 Conclusion

We have examined an understudied construction in East Javanese that has traditionally been analyzed as a passive, motivated by its optional by-phrase and typological similarities with true passives in neighboring Austronesian languages such as Indonesian and Acehnese (e.g. Cole et al. 2008; Kroeger 2014; Legate 2014). Six diagnostics instead show that the construction involves Ā-movement of the theme to a left-peripheral topic position, rather than A-movement to subject position, with the agent remaining syntactically active as the grammatical subject in a derived A-position. Its structural parallels and complementary person constraints with object voice further support a unified object-topicalization analysis, revealing that East Javanese exhibits a two-way, Ā-oriented voice system distinct from that of its neighboring languages. Together with the syntactic variation of the di-construction across Javanese dialects, these findings highlight the importance of distinguishing passivization from topicalization in languages without overt case marking and provide new evidence for a possible diachronic pathway linking Philippine-type and Indonesian-type voice systems.

The present analysis points to a potential diachronic pathway in which true passives develop from object topicalization through the reanalysis of topic-into-subject (Givón 1976; Comrie 1988; Shibatani 2011; inter alia). This grammaticalization pathway is attested in several language families (Indo-European, Pama-Nyungan, Japonic; see Shibatani 2011) and has recently been proposed for Western Austronesian languages (Patrianto & Chen 2023a; Lohninger & Katochoritis 2025). Further systematic investigation of the di-construction in other Javanese varieties and neighboring Austronesian languages, using the six diagnostics discussed in this paper (55), will help clarify this development.

    1. (55)
    1. Six diagnostics for the Ā vs. A-status of pivots in Indonesian-type passive constructions

Abbreviations

ACC Accusative MOD Modal
APPL Applicative NEG Negation
ATOP Aboutness topic NOM Nominative
AV Actor Voice OBL Oblique
CAUS Causative OV Object Voice
CLF Classifier PASS Passive
CM Common marker PERF Perfect
CN Common noun marker PIVOT Pivot
DEF Definite PL Plural
DEM Demonstrative PN.CM Proper noun common marker
FUT Future PN.PIVOT Proper noun pivot
GEN Genitive PREP Preposition
INDF Indefinite PROG Progressive
LK Linker RED Reduplication
REFL Reflexive

Funding information

This research was supported by a Marsden Grant (#MFP-VUW2012), a Victoria University of Wellington PhD scholarship, and a VUW Faculty Large Grant (#FGL-HSSE-10873).

Acknowledgements

We thank our language consultants for sharing their language with us: Awaludin Rusiandi, Bhakti Prasetya, Purnama Indra Cahyono, Mustafa Nur Fathoni, and Anang Santosa (Javanese); Maria Anunsiata M.I., Marie Angelique, Abdullah Sani, and Anita Bachtiar (Indonesian); and Murhaban, Cut Ida Agustina, Zainun, and Munzir (Acehnese). We are also grateful to Edith Aldridge, Michael Erlewine, Shin Fukuda, Jens Hopperdietzel, Brad McDonnell, Miriam Meyerhoff, Hiroki Nomoto, William O’Grady, Ileana Paul, Eric Potsdam, Maria Polinsky, and Jozina Vander Klok, as well as three anonymous Glossa reviewers and the audiences at NELS 53, AFLA 28–30 and the Austronesian Circle at University of Hawai‘i at Mānoa, for valuable feedback.

Competing interests

The authors have no competing interests to declare.

Notes

  1. We return to the syntactic status of these verbal prefixes in Sections 4 and 5. [^]
  2. Following recent work on Indonesian-type languages (Indonesian: Sukarno 2003; Acehnese: Ko 2009; Legate 2014; Sundanese: Kurniawan 2013; Balinese: Natarina 2018; Javanese: Patrianto & Chen 2023b; Cui & Rabinovitch 2024), we adopt the Voice/vP distinction throughout this paper and refer to verbal projections as VoicePs. We assume that an active Voice head introduces an external argument and assigns accusative case, whereas passive constructions involve a defective Voice head that is unable to license object case (Pylkkänen 2002; Harley 2013). In this sense, the VoiceP adopted here corresponds to the vP of earlier work (e.g. Aldridge 2008; Cole et al. 2008). This framework provides a more transparent basis for analyzing the passive-like construction examined here. [^]
  3. Note that the notion of pivot, as commonly adopted in more recent Austronesian literature, is often referred to as “subject” or “focused phrase” in earlier work on Javanese, such as Cole et al. (2002) and Davies (1993). These terms refer to the same underlying notion, as evidenced by the data and analyses presented in these works, including classic studies on Western Austronesian pivothood such as Schachter (1976) and De Guzman (1988). [^]
  4. Examples (19b) and (20a–b) come from primary fieldwork on Acehnese and Indonesian. The Indonesian data were collected from three monolingual speakers who do not speak other languages of Indonesia. See Arka & Manning (2008) for the same observation. [^]
  5. There is clear evidence that the reflexive awa’-é ḍéwé ‘himself/herself’ and its first- and second-person counterparts are genuine anaphors in East Javanese. Unlike pronouns and logophoric reflexives, they must be bound by a local antecedent. [^]
  6. We set aside Weak Crossover effects (Lasnik & Stowell 1991), which are not attested in our East Javanese data. [^]
  7. See also Davies (1993) for a detailed discussion of this construction. In that work, aboutness topics are represented as capitalized TOPIC, and their relationship to pivots is examined through evidence from possessor raising and extraction (relativization) constraints. The key point relevant here is that Javanese has two preverbal positions, both of which exhibit topic-like properties. Based on primary fieldwork and the first author’s native intuition, the higher topic is also associated with a distinct prosodic contour (including a pause), which does not apply to the pivot phrase. See Davies (1993) for further discussion of the interaction between these two positions. [^]
  8. We note Vander Klok (2019), which suggests that the topic-like behavior of Javanese pivots may reflect a pragmatic preference rather than a strict syntactic requirement, allowing pivots to remain in situ under certain conditions. The proposal is currently available as a conference abstract. The acceptability judgments we obtained from two East Javanese speakers from Surabaya differ from those reported there, and instead support the view that East Javanese pivots obligatorily move to a topic position. It is possible that the observations in Vander Klok (2019) are based on a different Javanese variety that does not enforce a strict topic constraint on pivots. For these reasons, we do not pursue a detailed comparison with that analysis here. [^]
  9. For analyses of the etymologically related actor-voice affix N- in Malay/Indonesian, see Saddy (1991), Soh (1998), Soh & Nomoto (2011), Aldridge (2008), and Erlewine & Sommerlot (2025). Note, however, that this line of analysis does not necessarily extend to East Javanese, as Malay/Indonesian has been reported to exhibit a subject-oriented voice system in which pivots behave as true grammatical subjects. [^]
  10. All five speakers consulted reported that the patterns in (38a–c) (each tested with five parallel sentences) were only marginally acceptable when pronounced with significantly raised intonation, yielding a focal reading and a partitive interpretation of the quantifier phrase. This contrasts with pre-auxiliary quantifier float in the AV sentence (4), which is judged more natural than its non-floating counterpart. [^]
  11. There is clear evidence that the default position of numeral quantifiers in East Javanese is pre-NP, as supported by noun phrase internal word order in aboutness topics, focus constructions, and NP modifiers, as shown in (i)–(ii).
      1. (i)
      1.  
      1. {
      2. {
      1. rong
      2. two
      1. buku
      2. book
      1. /
      2. /
      1. *buku
      2. book
      1. loro
      2. two
      1. },
      2. }
      1. ta’=Ø-kiro
      2. 1sg=ov-think
      1. Joko
      2. J
      1. wis
      2. already
      1. nyimpen
      2. av.keep/save
      1. wingi.
      2. yesterday
      1. ‘It was two books that I thought Joko kept yesterday.’
      1. (ii)
      1.  
      1. cerito-né
      2. story-def
      1. {
      2. {
      1. telung/telu
      2. three
      1. biḍaḍari
      2. female.angel
      1. /
      2. /
      1. *biḍaḍari
      2. female.angel
      1. telu
      2. three
      1. }
      2. }
      1. ‘the story of three angels’
    [^]
  12. We remain agnostic regarding the precise syntactic status of the person morpheme, which may be analyzed either as pronominal subject clitics or as subject agreement with a nominative DP. This issue is orthogonal to the central claim advanced here—that the morpheme indexes a nominative subject rather than an ergative agent, contrary to split-ergative analyses of Indonesian-type languages reviewed earlier. [^]
  13. We set aside the question of whether Javanese person-indexing affixes should be analyzed as pronominal clitics or agreement markers, as their status has not, to our knowledge, been comprehensively examined in the Austronesian literature. Crucially, the constraint observed in Javanese is not unique, but is better understood from a broader, pan-Austronesian perspective. [^]
  14. See, for example, Sneddon et al. (2010: 258–259) for a discussion of the third-person agent constraint in formal Indonesian passive constructions. Note that this pattern is not attested in many other varieties of Indonesian and other Malayic languages (see Nomoto 2020 for discussion). [^]
  15. It should be noted that we do not posit a direct derivational relationship between the Old Javanese case marker ni and Modern Javanese ambè’; rather, ambè’ appears to have replaced ni in marking non-pivot agents. The key generalisation is that non-pivot agents evolve from full NPs marked by a core case marker to prepositional phrases. [^]
  16. These observations align with previously reported data, although earlier descriptions do not explicitly state that pivots must be DPs or need not be specific. [^]
  17. The Indonesian variety used by monolingual Indonesian speakers is conventionally termed “Colloquial Indonesian” (e.g. Englebretson 2000). Given the multilingual context of Indonesia, “Indonesian monolingual” here refers to individuals who grew up [^]

References

Aldridge, Edith. 2008. Phase-based account of extraction in Indonesian. Lingua 118. 1440–1469. DOI:  http://doi.org/10.1016/j.lingua.2007.08.006

Andersen, Torben. 2015. Syntacticized topics in Kurmuk: A ternary voice-like system in Nilotic. Studies in Language 39. 508–54. DOI:  http://doi.org/10.1075/sl.39.3.01and

Arka, I. Wayan. 2008. Voice and being core: Evidence from (eastern) Indonesian languages. In Austin, Peter & Musgrave, Simon (eds.), Voice and grammatical relations in Indonesian: A new perspective. 183–227. Stanford: CSLI Publications.

Arka, I. Wayan & Manning, Christopher. 2008. Voice and grammatical relations in Indonesian: a new perspective. In Austin, Peter & Musgrave, Simon (eds.), Voice and grammatical relations in Indonesian: A new perspective, 45–69. Stanford: CSLI Publications.

Artawa, I. Ketut. 1998. Ergativity and Balinese syntax. Jakarta: Badan Penyelenggara Seri NUSA, Universitas Katolik Indonesia Atma Jaya.

Artawa, I. Ketut. 2013. The basic verb construction in Balinese. In Adelaar, Alexander (ed.), Voice variation in Austronesian languages of Indonesia. NUSA 54, 5-27.

Asikin-Garmager, Eli. 2017. Sasak voice. University of Iowa dissertation.

Badib, Abbas. 1980. Some major constructions in Javanese: A lexicase approach. Monash University dissertation.

Baier, Nicholas. 2018. Anti-agreement. UC Berkeley dissertation.

Bintoro. 1980. Javanese transitive verbs: A tagmemic grammar. Jakarta: Badan Penyelenggara Seri NUSA, Universitas Atma Jaya.

Blust, Robert. 2015. The case markers of Proto-Austronesian. Oceanic Linguistics 54(2). 436–91. DOI:  http://doi.org/10.1353/ol.2015.0024

Chen, Victoria. 2017. A reexamination of the Philippine-type voice system and its implications for Austronesian primary-level subgrouping. University of Hawai‘i at Mānoa dissertation.

Chen, Victoria. 2025. The syntax of Philippine-type alignment: Insights from case-marking. Natural Language & Linguistic Theory 43. 1839–1898. DOI:  http://doi.org/10.1007/s11049-025-09655-7

Chen, Victoria & McDonnell, Bradley. 2019. Western Austronesian voice. Annual Review of Linguistics 5(1). 173–195. DOI:  http://doi.org/10.1146/annurev-linguistics-011718-011731

Cole, Peter & Hermon, Gabriella & Inoha, Kozue & Tjung, Yassir. 2002. A constraint on wh in situ in Javanese. In Rackowski, Andrea & Richards, Norvin (eds.), Proceedings of AFLA 8, 91–106. MIT Working Papers in Linguistics 44.

Cole, Peter & Hermon, Gabriella & Yanti. 2008. Voice in Malay/Indonesian. Lingua 118. 1500–1553. DOI:  http://doi.org/10.1016/j.lingua.2007.08.008

Comrie, Bernard. 1988. Topics, grammaticalized topics, and subjects. In Axmaker, Shelley & Jaisser, Annie & Singmaster, Helen (eds.), Proceedings of the fourteenth annual meeting of the Berkeley Linguistics Society, 265–279. DOI:  http://doi.org/10.3765/bls.v14i0.1798

Conners, Thomas. 2008. Tengger Javanese. Yale University dissertation.

Cui, Jian & Rabinovitch, Jack Isaac. 2024. Two voices calling out as one: A split voice analysis of Javanese passives. Proceedings of the Linguistic Society of America 9(1). 5680–5680. DOI:  http://doi.org/10.3765/plsa.v9i1.5680

Davies, William. 1993. Javanese subjects and topics and psych verbs. Linguistics 31. 239–277. DOI:  http://doi.org/10.1515/ling.1993.31.2.239

De Guzman, Videa. 1988. Ergative analysis for Philippine languages: An analysis. In McGinn, Richard (ed.), Studies in Austronesian linguistics, 323–45. Ohio University Press.

Englebretson, Robert. 2000. Complementation in colloquial Indonesia conversation: A discourse-functional perspective. UC Santa Barbara dissertation.

Erlewine, Michael Yoshitaka & Sommerlot, Carly. 2025. Voice and extraction in Malayic. Language 101(4). 700–750. DOI:  http://doi.org/10.1353/lan.2025.a978273

Erlewine, Michael & Levin, Theodore & Van Urk, Coppe. 2019. The typology of nominal licensing in Austronesian voice system languages. In Paul, Ileana (ed.), Proceedings of the 26th meeting of Austronesian Formal Linguistics Association, 71–87.

Ewing, Michael. 2005. Grammar and inference in conversation: Identifying clause structure in spoken Javanese. Amsterdam: John Benjamins. DOI:  http://doi.org/10.1075/sidag.18

Fitzpatrick, Michael. 2006. The syntactic and semantic roots of floating quantification. MIT dissertation.

Givón, Talmy. 1976. Topic, pronoun and grammatical agreement. In Li, Charles N. (ed.), Subject and Topic, 151–188. New York: Academic Press

Harley, Heidi. 2013. External arguments and the mirror principle: On the distinctness of Voice and v. Lingua 125. 34–57. DOI:  http://doi.org/10.1016/j.lingua.2012.09.010

Hemmings, Charlotte. 2015. Kelabit voice: Philippine-type, Indonesian-type or something a bit different? Transactions of the Philological Society 113(3). 383–405. DOI:  http://doi.org/10.1111/1467-968X.12071

Hemmings, Charlotte. 2021. Subjects in Austronesian: Evidence from Kelabit. In Arka, I. Wayan & Asudeh, Ash & King, Tracy Holloway (eds.), Modular design of grammar, 160–180. Oxford: Oxford University Press. DOI:  http://doi.org/10.1093/oso/9780192844842.003.0011

Himmelmann, Nikolaus. 2002. Voice in Western Austronesian: An update. In Wouk, Fay & Ross, Malcolm (eds.), The history and typology of Western Austronesian voice systems, 7–15. Canberra: Pacific Linguistics.

Himmelmann, Nikolaus. 2020. Grammaticisation processes and reanalyses in Sulawesi languages. In Malchukov, Andre & Bisang, Walter (eds.), Areal patterns of grammaticalization and cross-linguistic variation in grammaticalization scenarios Band 2, 1043–1075. Berlin: de Gruyter. DOI:  http://doi.org/10.1515/9783110712735-011

Kaufman, Daniel. 2009. Austronesian typology and the nominalist hypothesis. In Adelaar, Alexander & Pawley, Andrew (eds.), Austronesian historical linguistics and culture history: A festschrift for Robert Blust, 179–226. Canberra: Australian National University Press.

Khairunnisa. 2022. Diathesis, grammatical relations, and clitics in Ampenan Sasak. University of Hawai‘i at Mānoa dissertation.

Ko, Seongyeon. 2009. Two types of anti causatives in Acehnese. In Chung, Sandy & Finer, Daniel & Paul, Ileana & Potsdam, Eric (eds.), The Proceedings of AFLA 16, 93–107.

Krauße, Daniel. 2017. A description of Surabayan Javanese: With special reference to its linguistic etiquette. Goethe University Master thesis.

Kroeger, Paul. 1993. Phrase structure and grammatical relations in Tagalog. Stanford, CA: Center for the Study of Language and Information, Stanford University.

Kroeger, Paul. 2014. Passive agents in Malay: The binding properties and discourse functions of agentive =nya. In Chung, Siaw-Fong & Nomoto, Hiroki (eds.), Current trends in Malay linguistics, 5–29. Tokyo: Tokyo University Foreign Studies.

Kurniawan, Eri. 2013. Sundanese complementation. University of Iowa dissertation.

Lambrecht, Knud. 1994. Information structure and sentence form: Topic, focus, and the mental representations of discourse referents. Cambridge University Press. DOI:  http://doi.org/10.1017/CBO9780511620607

Lasnik, Howard & Stowell, Tim. 1991. Weakest crossover. Linguistic Inquiry 22(4). 687–720.

Legate, Julie. 2012. Subjects in Acehnese and the nature of the passive. Language 88(3). 495–525. DOI:  http://doi.org/10.1353/lan.2012.0069

Legate, Julie. 2014. Voice and v: Lessons from Achenese. MIT Press. DOI:  http://doi.org/10.7551/mitpress/9780262028141.001.0001

Lohninger, Magdalena & Katochoritis, Yiannis. 2025. Between topics and subjects: An Ā/A typology of Austronesian pivots. In Akolkar, Shweta & Galvano, Amber & Ismael, Akil & Liu, Kang Franco & Mikkelsen, Line (eds.) Proceedings of WCCFL 42, 224–231. Somerville, MA: Cascadilla Proceedings Project.

McDonnell, Bradley & Chen, Victoria. 2022. The evolution of syntax in Western Austronesian. In Shei, Chris & Li, Saihong (eds.), The Routledge handbook of Asian linguistics, 11–32. Routledge. DOI:  http://doi.org/10.4324/9781003090205-3

Miyagawa, Shigeru. 2010. Why agree? Why move? Cambridge, MA: MIT Press. DOI:  http://doi.org/10.7551/mitpress/8116.001.0001

Miyagawa, Shigeru. 1989. Structure and case marking in Japanese. San Diego, CA: Academic Press. DOI:  http://doi.org/10.1163/9789004373259

Nagaya, Naonori. 2012. The Lamaholot language of eastern Indonesia. Rice University dissertation.

Natarina, Ari. 2018. Complementation in Balinese: Typological, syntactic, and cognitive perspectives. University of Iowa dissertation.

Nomoto, Hiroki. 2018. The development of the English-type passive in Balinese. Wacana, Journal of the Humanities of Indonesia 19(1). Article 6. DOI:  http://doi.org/10.17510/wacana.v19i1.620

Nomoto, Hiroki. 2020. Bare passive agent hierarchy. In Hsieh, Henrison & New, Keely (eds.), Proceedings of the Austronesian Formal Linguistics Association (AFLA) 27, 57–70.

Nurhayani, Ika. 2014. A unified account of the syntax of valence in Javanese. Cornell University dissertation.

Ouhalla, Jamal. 1993. Subject-extraction, negation and the antiagreement effect. Natural Language & Linguistic Theory 11(3). 477–518. DOI:  http://doi.org/10.1007/BF00993167

Ogloblin, Alexander. 2005. Javanese. In Adelaar, Alexander & Himmelmann, Nikolaus (eds.), The Austronesian languages of Asia and Madagascar. London: Routledge.

Patrianto, Hero & Chen, Victoria. 2023a. A cline of Indonesian-type voice as in transition from Ā- to A-syntax: Insights from four languages. Talk given at the 30th annual meeting of the Austronesian Formal Linguistic Association. Lund University, October 28–31.

Patrianto, Hero, & Chen, Victoria. 2023b. Rethinking Indonesian-type passives: Lessons from Javanese. In Lam, Suet-Ying & Ozaki, Satoru (eds.), Proceedings of the 53rd annual meeting of the North East Linguistic Society (NELS 53), 45–58. Amherst, MA: GSLA.

Pearson, Matthew. 2005. The Malagasy subject/topic as an Ā-element. Natural Language and Linguistic Theory 23. 381–457. DOI:  http://doi.org/10.1007/s11049-004-1582-7

Poedjosoedarmo, Gloria. 1977. Thematization and information structure in Javanese. In Halim, Amran (ed.), Miscellaneous studies in Indonesian and languages in Indonesia, 34–43. Part II. Jakarta: Badan Penyelenggara Seri NUSA.

Poedjosoedarmo, Gloria. 1986. Role structure in Javanese. Jakarta: Badan Penerbit NUSA, Universitas Katolik Indonesia Atmajaya.

Poedjosoedarmo, Gloria. 2002. Changes in word order and noun phrase marking from Old to modern Javanese: Implications for understanding developments in Western Austronesian “focus” systems. In Wouk, Fay & Ross, Malcolm (eds.), The history and typology of western Austronesian voice systems, 311–330. Canberra: Australian National University Press.

Prince, Ellen. 1981. Topicalization, focus-movement, and Yiddish-movement: A pragmatic differentiation. Proceedings of the seventh annual meeting of the Berkeley Linguistics Society, 249–264.

Pylkkänen, Liina. 2002. Introducing arguments. MIT dissertation.

Rackowski, Andrea. 2002. The structure of Tagalog: Specificity, voice, and the distribution of arguments. MIT dissertation.

Rackowski, Andrea & Richards, Norvin. 2005. Phase edge and extraction: A Tagalog case study. Linguistic Inquiry 36(4). 565–599. DOI:  http://doi.org/10.1162/002438905774464368

Ramelan. 1983. Javanese indicative and imperative passives. In Halim, Amran & Carrington, Lois & Wurm, Stephen A. (eds.), Papers from the third international conference on Austronesian Linguistics Vol. 4: Thematic variation, 199–214. Pacific Linguistics, The Australian National University.

Ras, Johannes. 1994. History writing and the legitimacy of Javanese kingship. Bijdragen tot de taal-, land- en volkenkunde 150(3). 518–538. DOI:  http://doi.org/10.1163/22134379-90003075

Richards, Norvin. 2000. Another look at Tagalog subjects. In Paul, Ileana & Phillips, Vivianne & Travis, Lisa (eds.), Formal issues in Austronesian linguistics, 105–16. Dordrecht: Kluwer. DOI:  http://doi.org/10.1007/978-94-017-1580-5_6

Rizzi, Luigi. 1990. Relativized minimality. Cambridge, MA: MIT Press.

Rizzi, Luigi. 1997. The fine structure of the left periphery. In Haegeman, Liliane (ed.), Elements of grammar, 81–337. Berkeley: Kluwer. DOI:  http://doi.org/10.1007/978-94-011-5420-8_7

Robson, Stuart. 2014. Javanese grammar for students. Monash University Publishing.

Ross, Malcolm. 2002. The history of Western Austronesian voice and voice-marking. In Wouk, Fay & Ross, Malcolm (eds.), The history and typology of Western Austronesian voice systems, 17–62. Canberra: Pacific Linguistics.

Ross, Malcolm. 2009. Proto Austronesian verbal morphology: a reappraisal. In Adelaar, Alexander & Pawley, Andrew (eds.), Austronesian Historical Linguistics and Culture History: A Festschrift for Robert Blust 295–326. Canberra: Pacific Linguistics.

Saddy, Douglas. 1991. Wh-scope mechanisms in Bahasa Indonesia. In Cheng, Lisa & Demirdash, Hamida (eds.), MIT working papers in linguistics 15: More papers on wh-movement, 183–218.

Sato, Yosuke. 2015. Argument ellipsis in Javanese and voice agreement. Studia Linguistica 69(1). 58–85. DOI:  http://doi.org/10.1111/stul.12029

Schachter, Paul. 1976. The subject in Philippine languages: Topic, actor, actor-topic, or none of the above? In Li, Charles N. (ed.), Subject and topic, 491–518.

Shibatani, Masayoshi. 1985. Passives and related constructions: A prototype analysis. Language 61. 821–848. DOI:  http://doi.org/10.2307/414491

Shibatani, Masayoshi. 1988. Voice in Philippine languages. In Shibatani, Masayoshi (ed.), Passive and voice, 85–142. Amsterdam: John Benjamins. DOI:  http://doi.org/10.1075/tsl.16.06shi

Shibatani, Masayoshi. 2011. Grammaticization of topic into subject. In Traugott, Elizabeth & Heine, Bernd (eds.), Approaches to grammaticalization, 93–134. John Benjamins Publishing Company.

Sirima, Aarin. 2025. Voice in transition: A diachronic-typological study of Mori Atas. The Australian National University dissertation.

Sneddon, James & Adelaar, Alexander & Djenar, Dwi Noverini & Ewing, Michael. 2010. Indonesian reference grammar. Second edition. Crows Nest, N.S.W.: Allen & Unwin.

Sofwan, Ahmad. 2001. Javanese passives, ergatives and adversatives. La Trobe Papers in Linguistics 11(8).

Soh, Hooi Ling. 1998. Certain restrictions on A-bar movement in Malay. In Pearson, Matthew (ed.), Proceedings of the third and fourth meetings of Austronesian Linguistics Association 1996–1999, 295–308.

Soh, Hooi Ling & Nomoto, Hiroki. 2011. The Malay verbal prefix meN- and the unergative/unaccusative distinction. Journal of East Asian Linguistics 20(1). 77–106. DOI:  http://doi.org/10.1007/s10831-010-9069-5

Sportiche, Dominique. 1988. A theory of floating quantifiers and its corollaries for constituent structure. Linguistic Inquiry 19(3). 425–449.

Suhandano, Ignatius. 1994. Grammatical relations in Javanese: A short description. Australian National University MA thesis.

Suharno, Ignatius. 1982. A descriptive study of Javanese. Department of Linguistics, Research School of Pacific Studies, The Australian National University.

Sukarno, Wahyono. 2003. Derivational syntax: A minimalist approach to affixation in Bahasa Indonesia predicates. Victoria University of Wellington dissertation.

Teng, Stacy Fang-ching. 2009. Case syncretism in Puyuma. Language and Linguistics 10(4). 819–844.

Uhlenbeck, Eugenius M. 1975. Sentence segment and word group: Basic concepts of Javanese syntax. In Verhaar, John W. M. (ed.), NUSA 1, 6–10. Jakarta: NUSA.

van Urk, Coppe. 2015. A uniform syntax for phrasal movement: A case study of Dinka Bor. MIT dissertation.

Van den Berg, Rene. 2004. Some notes on the origin of Malay di. Bijdragen tot de Taal-, Land-en Volkenkunde 160(4). 532–554. DOI:  http://doi.org/10.1163/22134379-90003722

Vander Klok. 2008. Javanese modals. In Jones, Susie (ed.), Proceedings of the 2008 annual conference of the Canadian Linguistic Association, 1–15.

Vander Klok. 2019. Distinguishing subjects and topics in Javanese. Talk given at the 26th annual meeting of the Austronesian Formal Linguistic Association. University of Western Ontario, May 24–26.

Wechsler, Stephen & Arka, I. Wayan 1998. Syntactic ergativity in Balinese: An argument structure based theory. Natural Language & Linguistic Theory 16(2). 387–442. DOI:  http://doi.org/10.1023/A:1005920831550

Wedhawati & Arifin, Sjamsul. 2006. Tata bahasa Jawa mutakhir. Penerbit Kanisius: Yogyakarta.