In this chapter, results of the typological survey are considered in greater detail through the lens of geographic typology. For each geographic subregion, a closer look is taken at the distribution of applicatives vis-a-vis the typological profile of languages in that subregion. These patterns are used to interpret the distributional results presented in the previous chapter, especially the reasons why some West Nusantara languages lack applicatives, others retained pivot-selecting applicatives, and some developed pivot-neutral applicatives.
This chapter is organized geographically, treating in turn the languages of Sumatra and the Barrier Islands (§ 51), Java and Madura (§ 52), the Lesser Sundas (§ 53), Mainland Southeast Asia and Peninsular Malaysia (§ 54), Borneo and the Southern Philippines (§ 55), and Sulawesi (§ 56). A summary of findings is given in § 57.
51 Sumatra and the Barrier Islands
Sumatra represents the westernmost of the major islands in West Nusantara, and the Barrier Island chain runs from north to south off its west coast. The languages of Sumatra and the Barrier Islands are diverse with respect to genetic affiliation within Malayo-Polynesian. They include (i) Malay languages (i.e. the Malay subbranch of Malayic), (ii) the Northwest Sumatra-Barrier Islands group, (iii) Acehnese, which belongs to the Chamic group, and (iv) some small outlier groupings and isolates including Enggano, Nasal, Rejang, and Lampungic.
As shown in Table 19, 13 languages of Sumatra and the Barrier Islands are included in the sample. Due to lack of adequate descriptive material, languages of the Barrier Islands are not well represented in the sample. To address this, in this section, when partial information is available I will make mention of two additional languages spoken in the Barrier Islands: Mentawai and Sigulai (Sikule) (Adriani 1928; Morris 1900; Kähler 1955; see also McDonnell & Truong 2024b). Both belong to the Northwest Sumatra-Barrier Islands group. An overview map of the languages of the sample in this geographic area and other languages discussed in this section is presented in Map 4.



Map 4
Language overview for Sumatra and the Barrier Islands
Table 19
Typological survey results for Sumatra and the Barrier Islands
|
Gen. grp. |
Name |
Applicatives? |
Voice |
Morph. align. |
Word order |
|
|---|---|---|---|---|---|---|
|
Pivot- |
Pivot- |
|||||
|
selecting |
neutral |
|||||
|
Chamic |
Acehnese |
N |
Y |
asymmetrical |
split-S |
free |
|
nws-bi |
Alas Batak |
N |
Y |
two-way |
mixed-npiv.a |
avp |
|
nws-bi |
Karo Batak |
N |
Y |
two-way |
pivot-nonpivot |
avp |
|
nws-bi |
Toba Batak |
N |
Y |
two-way |
mixed-npiv.a |
vpa |
|
nws-bi |
Gayo |
N |
Y |
two-way |
mixed-npiv.a |
avp |
|
nws-bi |
Nias |
N |
Y |
marg. two-way |
mixed:other |
vpa |
|
Enggano |
Enggano |
N |
Y |
asymmetrical |
neutral |
avp |
|
Nasal |
Nasal |
Y |
Y |
two-way |
pivot-nonpivot |
avp |
|
Malayic |
Kerinci Malay |
N |
Y |
two-way |
mixed:other |
avp |
|
Malayic |
Jambi Malay |
N |
N |
two-way |
mixed:other |
avp |
|
Malayic |
S. Barisan Malay |
N |
Y |
two-way |
mixed-npiv.a |
avp/vpa |
|
Rejang |
Rejang |
N |
N |
two-way |
mixed-npiv.a |
avp |
|
Lampungic |
Lampung Api |
N |
Y |
two-way |
mixed-npiv.a |
avp |
Across all genetic groupings in Sumatra, the presence of pivot-neutral applicatives is the norm (11 of 13 languages of the sample),1 and pivot-neutral applicatives are also found in Mentawai and Sigulai. Pivot-selecting applicatives are not found in Sumatra and the Barrier Islands (0 of 13), and there is no clear evidence that they can be reconstructed for any of the genetic groupings represented. Sporadic cases of lack of applicatives are found in two languages (2 of 13); Rejang and Kerinci do not have applicatives of any type (see § 51.1 below).
Pivot-neutral applicatives are found across languages with diverse types of voice systems in Sumatra and the Barrier Islands. They are found in languages with asymmetrical voice systems (2 of 2; Acehnese, Enggano) and one marginal two-way voice system (1 of 1; Nias),2 as well as the large majority of two-way symmetrical voice systems (8 of 10), the most common type of voice system in languages spoken on the island of Sumatra.3
Pivot-neutral applicatives are found across languages with diverse systems of morphological alignment as well. Quite common are systems of morphological alignment in which non-pivot A arguments have distinct forms compared to all other core arguments, especially for languages spoken on the island of Sumatra. In the sample, almost all such languages (5 of 6) show pivot-neutral applicatives. The remaining seven languages in the sample are split across four coding categories for morphological alignment: split intransitive (1), pivot-non-pivot (2), neutral (1), and mixed (other) (3). Nonetheless almost all of these languages (6 of 7) also have pivot-neutral applicatives.
The majority of languages of Sumatra and the Barrier Islands have avp unmarked word order in A-oriented transitive constructions, with most of these (7 of 9) also showing pivot-neutral applicatives. Verb-initial word order is preferred in Nias and Toba Batak, which both have pivot-neutral applicatives (2 of 2), and this also appears to hold for Sigulai (see McDonnell & Truong 2024b: 422). In Acehnese, word order is relatively free, with the majority of clauses being verb-initial (Durie 1985: 191). In South Barisan Malay, the non-pivot immediately follows the verb, while the pivot has freer word order (avp/vpa in av). Both Acehnese and South Barisan Malay have pivot-neutral applicatives.
On the whole, it is clear that the presence of pivot-neutral applicatives is the norm in Sumatra and the Barrier Islands, and these constructions show a broad distribution, being found across all five genetic groupings, all four types of voice systems, and all five coding categories for morphological alignment represented, as well as both languages with verb-initial unmarked word order and those with verb-medial word order. Survey results for Sumatra and the Barrier Islands thus support the finding that the presence of pivot-neutral applicatives is not associated with a particular typological profile within West Nusantara.
In particular, the presence of pivot-neutral applicatives in Mentawai and Enggano, is telling. Mentawai has two pivot-neutral applicative suffixes, ‑ake, which forms instrument- and theme-selecting applicative constructions (ACs), and ‑i, which forms locative- and goal-selecting ACs (Adriani 1928: 69–70, 81–83). Mentawai speakers, as inhabitants of Siberut, Sipura, and the Pagai islands west of Sumatra, for many centuries had minimal contact with outsiders prior to first visits by British and Dutch colonists in the late 1700s. There is little sign of prior Indic or Islamic cultural influence in the Mentawai islands (Loeb 1928), including that of the powerful Malay-speaking kingdoms established in western Indonesia and Malaysia from the 7th century until the 16th century. Therefore, it is highly unlikely that applicatives were introduced to Mentawai via language contact.
Enggano, spoken on the southernmost of the Barrier Islands, is considered one of the most aberrant Austronesian languages due to its unusual lexicon, and has even been argued to be non-Austronesian though it is now fairly well-established that it belongs to Malayo-Polynesian (Edwards 2012). Enggano also shows unusual characteristics for West Nusantara in its grammatical system, including its system of case and alignment.4 Enggano makes use of nominal prefixes to distinguish core arguments from obliques (neutral alignment) and the forms of these prefixes co-vary with grammatical number, human vs. nonhuman referent, and status as a proper vs. common noun (Crowley n.d.). Enggano also has two pivot-neutral applicative suffixes, which are illustrated in examples (101) and (102).
(101) Enggano, Locative-selecting applicative
a.
kia
ki-hɛ̃kũ
i-kuɔ
eʔana.
3sg
verb-sit
loc.cn-tree
med
‘He is sitting in that tree.’ (BC)
b.
kia
ki-hɛ̃kũ-hũĩ
e-kuɔ
eʔana.
3sg
verb-sit-loc.appl
core.cn-tree
med
‘He is sitting in that tree.’ (AC)(Kähler 1940: 195, English translation & glosses added)
(102) Enggano, Instrument-selecting applicative
a.
kia
ki-pudu
e-kɔyɔ
iʔiɔː
u-bɔhɛ
3sg
verb-kill
core.cn-pig
loc.prep
obl.cn-spear
‘He killed the wild pig with a spear.’ (BC)
b.
kia
ki-pudu-ʔa
e-bɔhɛ
iʔiɔː
u-kɔyɔ
3sg
verb-kill-inst.appl
core.cn-spear
loc.prep
obl.cn-pig
‘He killed the wild pig with a spear.’ (AC)(Kähler 1940: 196, English translation & glosses added)
Edwards (2012) offers two possible explanations for the aberrant nature of Enggano: “(i) that Enggano suffered contact effects with a non-Austronesian language present in the region prior to the appearance of [Malayo-Polynesian] languages, and/or (ii) that Enggano Island was relatively isolated from the rest of the Austronesian world.” If the latter case is true, the presence of pivot-neutral applicatives in Enggano further bolsters the argument that the distribution of pivot-neutral applicatives in West Nusantara is not primarily explained by language contact with culturally prominent languages. But even if it is not, Enggano’s unusual case markers and system of alignment still underscore another major finding of this chapter, that the presence of pivot-neutral applicatives is not associated with a particular typological profile within West Nusantara.
51.1 Outliers in Sumatra
In Sumatra and the Barrier Islands, only Kerinci (Malayic) and Rejang (isolate) lack applicatives completely. The other 11 languages of the sample—plus Mentawai and Sigulai—have pivot-neutral applicatives and these are almost exclusively marked with verbal suffixes. Only Acehnese has an AC marked with a prefix, i.e. peu‑, and this is of limited productivity.
Unlike other languages of Sumatra and the Barrier Islands, Rejang lacks suffixes of any type and uses only prefixes (McGinn 1982). This is likely the reason for the lack of applicatives in Rejang. However, we do not have evidence by which to determine whether (i) Rejang never had applicative suffixes, or (ii) some such suffixes were previously present but later lost (cf. the case of Mualang, where clausal alternations persist after the loss of applicative suffixes, see § 55.5). Rejang is currently classified as an isolate within Malayo-Polynesian, and not much is known about its pre-history, including whether the language arrived in Sumatra after a period of development elsewhere in West Nusantara, such as Borneo (see McGinn 2009). Therefore, I consider this an open question, awaiting more historical evidence linking Rejang genetically or geographically with other languages of West Nusantara.
In Kerinci, it was not loss of suffixation per se, but phonological changes involving fusion of certain suffixes (and pronominal enclitics) to stems, that have resulted in a lack of applicatives (Yanti et al. 2018). It is clear that pivot-neutral applicative suffixes were among the morphemes lost through fusion, as one such suffix, cognate with Standard Malay/Indonesian ‑kan, is found in remnant forms in the Tanjung Pauh variety of Kerinci, e.g. kato-ka ‘say (s.t.)’ (Yanti et al. 2018: 469). This process of fusion has resulted in radical changes in the grammar of Kerinci that have been of some interest in the literature, including the genesis of “absolute” and “oblique” forms of stems that color its unusual system of agreement and mixed system of morphological alignment (McKinnon, Cole & Hermon 2011; Yanti 2010).
52 Java and Madura
Java represents the most populous of the major islands in West Nusantara, and Madura, a smaller island off the northeast coast of Java. These islands are home to the Sundic languages, Javanese languages, and Madurese languages (see Map 5). Javanese languages are also found outside of West Nusantara in New Caledonia (Pacific region), and Suriname (a Caribbean nation of South America, see Map 6), where they are spoken by diaspora communities of ethnically Javanese people.
The survey sample includes one Sundic language, three Javanese languages, and one Madurese language, as shown in Table 20, which includes Standard Indonesian as well for reference. These languages share a similar typological profile. They have two-way symmetrical voice systems with pivot-neutral applicatives that are marked with verbal suffixes. In unmarked word order, the non-pivot argument immediately follows the verb, and the pivot argument is preverbal. Thus most have predominant avp word order in av, with vpa also possible. These languages have one distinct morphological marking for the non-pivot A argument and one for all other core grammatical relations. The type of marking may be dependent on the person category of the non-pivot A argument. In Sundanese and Madurese, the verb in a pv construction is marked with a voice prefix, and the non-pivot A argument is marked with a preposition meaning ‘with, by’ (i.e. ku in Sundanese and bi’ in Madurese). The preposition is optional if A immediately follows the verb as in unmarked word order, but obligatory in other word orders. In Javanese languages, third person non-pivot A arguments follow the same pattern for optional prepositional marking, while first and second person non-pivot A arguments are expressed with special proclitic pronominal forms that replace the pv voice prefix.5



Map 5
Language overview for Java, Madura, and the Lesser Sundas



Table 20
Typological survey results for Java and Madura
|
Gen. grp. |
Name |
Applicatives? |
Voice |
Morph. align. |
Word order |
|
|---|---|---|---|---|---|---|
|
Pivot- |
Pivot- |
|||||
|
selecting |
neutral |
|||||
|
Sundic |
Sundanese |
N |
Y |
two-way |
mixed-npiv.a |
avp |
|
Javanese |
Javanese |
N |
Y |
two-way |
mixed-npiv.a |
avp |
|
Javanese |
Tengger |
N |
Y |
two-way |
mixed-npiv.a |
avp |
|
Javanese |
Sur. Javanese |
N |
Y |
two-way |
mixed-npiv.a |
avp |
|
Madurese |
Madurese |
N |
Y |
two-way |
mixed-npiv.a |
avp/vpa |
|
Malayic |
Std. Indonesian |
N |
Y |
two-way |
mixed-npiv.a |
avp |
Also spoken on Java but not included in the survey sample are Betawi Malay and Colloquial Jakarta Indonesian, both spoken in and around the special administrative region of Jakarta, the capital of Indonesia. Betawi is the language spoken by the original inhabitants of the Jakarta area. It is not included in the survey due to its classification as a Malay-based creole by Eberhard, Simons & Fennig (2021). Colloquial Jakarta Indonesian is a variety of standard Indonesian influenced by Betawi. It is not included in the survey because it does not have a separate iso-639-3 code, but shares one with Standard Indonesian. Like other languages of Java, both Betawi and Colloquial Jakarta Indonesian have two-way symmetrical voice systems and pivot-neutral applicatives marked with verbal suffixes (Ikranagara 1975; Sneddon 2006: 30–34).
Overall, the languages of Java and Madura show a fairly coherent typological profile, one that is also associated with Standard Indonesian and that is consistent with the defining features of the proposed Indonesian-type of western Austronesian languages. However, when viewed in light of the full results of the typological survey, this profile is too limited and does not explain the full distribution of pivot-neutral applicatives in West Nusantara.
53 The Lesser Sundas
The Lesser Sundas are a chain of islands stretching eastward from the east coast of Java (see Map 5). In the Lesser Sundas, only the westernmost islands of Bali and Lombok are considered to constitute part of West Nusantara, with Sumbawa, Sumba, Flores, and Timor generally being defined as part of East Nusantara (see Klamer & Ewing 2010). But because Balinese, spoken on Bali, Sasak, spoken on Lombok, and Sumbawa, spoken on Sumbawa Island, together form a well-defined lower-level subgroup of Malayo-Polynesian, all three are eligible for inclusion in the survey. Of the three languages, Sasak and Sumbawa form one primary branch, while Balinese represents another. As shown in Table 21, all three languages are included in the sample. In the case of Sasak, survey data was compiled based on the Ampenan variety (Khairunnisa 2022), which shows some differences with other Sasak varieties described by Austin (2001).
Though they represent a small, closely-related genetic grouping, Balinese, Sasak, and Sumbawa show important differences in their typological profiles. These languages show a cline of features, with Balinese being typologically similar to the languages of Java and Madura, Sumbawa being similar to other East Nusantara languages, and Sasak occupying a intermediate position.
Table 21
Typological survey results for the Lesser Sundas
|
Gen. grp. |
Name |
Applicatives? |
Voice |
Morph. align. |
Word order |
|
|---|---|---|---|---|---|---|
|
Pivot- |
Pivot- |
|||||
|
selecting |
neutral |
|||||
|
Bali-Sasak-Sumbawa |
Balinese |
N |
Y |
two-way |
mixed-npiv.a |
avp |
|
Bali-Sasak-Sumbawa |
Ampenan Sasak |
N |
Y |
two-way |
mixed:other |
avp |
|
Bali-Sasak-Sumbawa |
Sumbawa |
N |
N |
asymmetrical |
accusative |
undet. |
The voice and applicative systems of Balinese were discussed in § 34 above. The language shows a two-way symmetrical voice system, with av marked with the prefix N‑, as well as pivot-neutral applicatives marked with two suffixal forms. In Balinese, third person non-pivot A arguments have a special morphological form that distinguishes them from other core arguments; these arguments may be expressed with the enclitic =na. In all other cases, core arguments are realized as bare NPs or full pronouns, when expressed. Normally, the pivot precedes the verb followed by the non-pivot argument; accordingly, the unmarked word order in av is categorized as avp. Balinese shows both prefixes and suffixes, with a relatively small number of total affixes (Shiohara & Arka 2024).
Like Balinese, Ampenan Sasak contrasts A-oriented and P-oriented transitive constructions, however the distinction between them is usually not morphologically marked on the verb (Khairunnisa 2022). Consequently, Ampenan Sasak may be considered a two-way symmetrical diathesis system. Like Balinese, Ampenan Sasak also shows pivot-neutral applicatives, but only one suffixal form is used to mark these. Ampenan Sasak shows a complex, mixed system of morphological alignment. In P-Diathesis, a non-pivot A argument expressed as an NP has special marking with siq. However, the use of clitic pronouns, which are extremely common, shows a different pattern of alignment: P arguments expressed as clitics must be enclitics regardless of diathetical construction, while A arguments expressed as clitics may be either proclitics or enclitics. In canonical word order, the pivot is preverbal; accordingly, the unmarked word order in A-oriented constructions is avp. While Ampenan has a number of prefixes, its sole suffix is ‑an, which is highly polyfunctional, and is used to mark the verb in ACs (Khairunnisa & McDonnell in prep).
Sumbawa shows even greater differences with Balinese while sharing some characteristics with Ampenan Sasak. Sumbawa shows an asymmetrical diathesis system, and has “only one type of transitive construction with invariably bare verbs” (Shiohara & Arka 2024: 500). It does not have applicatives of any type. It shows accusative morphological alignment, with pronominal proclitics expressing the S and A core relations, but not P.6 In these respects, Sumbawa resembles other Austronesian languages of East Nusantara, which do not show morphological marking for voice, and make use of proclitics or prefixes indexing the clausal subject or agent (Klamer & Ewing 2010: 10). Many East Nusantara languages also do not have applicatives (though a few do, e.g. Taba, Bowden 2001). There is no identified predominant word order for transitive clauses in Sumbawa, however verb-initial clauses are most common in narrative texts (Shiohara & Arka 2024). Sumbawa shows only prefixes and no suffixes.
Bali-Sasak-Sumbawa represents a transition zone standing between the languages of Java and Madura to the west and East Nusantara languages to the east (Klamer & Ewing 2010). Balinese has the most verbal morphology including av prefixal marking and two applicatives suffixes. Ampenan Sasak shows reduced morphological marking for voice and one applicative suffix. Sumbawa has the least verbal morphology and no applicatives. Further evidence of this transition is observed across Sasak dialects. Like Balinese, Ngenó-Ngené Sasak has two applicative morphemes (AMs), ‑ang and ‑in, and contrasts A-oriented and P-oriented constructions (Austin 2001). Menó-Mené Sasak has just ‑ang, making its applicative system similar to Ampenan Sasak, but its voice system is more like Sumbawa, as Menó-Mené does not show evidence of symmetrical alternations in voice or diathesis.
Because pivot-neutral applicatives are found in both primary branches of the subgroup, they can be reconstructed to Proto Bali-Sasak-Sumbawa. Pivot-neutral applicatives are unlikely to have been borrowed wholesale from Balinese into Sasak, as the AM suffix *‑an, reflected in Balinese, Ngenó-Ngené, and Menó-Mené as ‑ang (Austin 2001) and Ampenan Sasak as ‑an, is quite old. It is also found with applicative functions in Proto Malayic, which has been proposed to subgroup with Chamic and then Bali-Sasak-Sumbawa at some higher level under Malayo-Polynesian (Adelaar 2005a). Most likely then, Sumbawa represents a case of loss of pivot-neutral applicatives, apparently due to loss of suffixation and under the influence of neighboring East Nusantara languages.
54 Mainland Southeast Asia and Peninsular Malaysia
In this section, I present and discuss results of the typological survey for mainland Southeast Asia and Peninsular Malaysia. This geographic area includes (i) the Malay peninsula, which is considered part of West Nusantara and politically belongs to Malaysia, and (ii) other parts of Southeast Asia to the north of the Malay peninsula, which are not part of West Nusantara proper, and are governed by Myanmar, Thailand, Laos, Cambodia, Vietnam, and China (Hainan Island). Austronesian languages spoken in mainland Southeast Asia and Peninsular Malaysia include Chamic languages, Malay languages, and the Moken-Moklen languages, which are not included in the sample, but are discussed as a point of comparison in this section. Map 7 shows languages of the sample in this geographic area and other languages and varieties discussed in this section. A summary of results for languages of the sample is given in Table 22.



Map 7
Language overview for mainland Southeast Asia and Peninsular Malaysia
Chamic is made up of 12 total languages, with four included in the sample. One of these, Acehnese, is spoken in Aceh Province in northern Sumatra, and is discussed in § 51 above, while the other three, Eastern Cham, Bih, and Tsat, are spoken in mainland Southeast Asia, and are discussed in this section. Within Malayo-Polynesian, Malayic languages are the most closely related to Chamic and are considered “next-of-kin” (Thurgood 1999).
Malay is a primary branch of the larger Malayic genetic group. Of the 42 Malayic languages, 33 are classified as Malay. Two Malay languages are spoken in mainland Southeast Asia and both are included in the sample: Pattani Malay and Urak Lawoi’. In addition, about nine Malay languages are spoken on the Malay peninsula, however, for many of these, there is little available grammatical description, aside from studies of phonological differences with Standard Malay. Four Malay languages and varieties not included in the sample will be discussed in this section on the basis of available information: Kedah Malay, Negeri Sembilan Malay, and the Nonthaburi and Kelantan varieties (which share one iso-639-3 code with Pattani Malay). Standard Malay, the national language of Malaysia, is not included in the sample because of its similarities to Standard Indonesian, but will also be discussed here. Singapore Bazaar Malay, a Malay language used in interethnic communication in Singapore, just to the south of the Malay peninsula, is included in the sample (but was coded as “other” for location), and will be discussed in this section as a point of comparison.
Table 22
Typological survey results for mainland SE Asia and Peninsular Malaysia
|
Gen. grp. |
Name |
Applicatives? |
Voice |
Morph. align. |
Word order |
|
|---|---|---|---|---|---|---|
|
Pivot- |
Pivot- |
|||||
|
selecting |
neutral |
|||||
|
Chamic |
Eastern Cham |
N |
N |
asymmetrical |
neutral |
avp |
|
Chamic |
Bih |
N |
N |
asymmetrical |
neutral |
avp |
|
Chamic |
Tsat |
N |
N |
asymmetrical |
neutral |
avp |
|
Malayic |
Urak Lawoi’ |
N |
N |
asymmetrical |
neutral |
avp |
|
Malayic |
Pattani Malay |
N |
N |
two-way† |
mixed-npiv.a |
avp |
|
Malayic |
Sing. Bazaar Malay |
N |
N |
asymmetrical |
neutral |
avp |
† The Nonthaburi variety of Pattani Malay shows an asymmetrical voice system.
The Austronesian languages of mainland Southeast Asia and northern Peninsular Malaysia show a strong tendency to lack applicatives altogether. Chamic languages spoken in mainland Southeast Asia all lack applicatives (3 of 3), as do Moken and Moklen (Larish 2005). Malayic languages spoken in mainland Southeast Asia (2 of 2; Pattani, Urak Lawoi’) again lack applicatives, and this holds also for the Nonthaburi variety spoken in Central Thailand. In northern Peninsular Malaysia, Kedah Malay lacks applicatives (Omar 1981: 7), and Kelantan Malay shows only cases of fossilized applicative suffixes (Wu 2023).
Moving further southward, Negeri Sembilan Malay, spoken in southern Peninsular Malaysia, has productive pivot-neutral ACs formed with both ‑i and ‑kan (Hendon 1966: 61–68), just like Standard Malay and Indonesian. On the other hand, Singapore Bazaar Malay lacks applicatives altogether.
Austronesian languages of mainland Southeast Asia and Peninsular Malaysia that lack applicatives tend to show typological features different from most other languages of West Nusantara. Chamic languages have asymmetrical voice systems (3 of 3), do not show morphological marking that distinguishes core arguments by grammatical relation (neutral alignment, 3 of 3), and show avp word order in transitive constructions (3 of 3), with word order being important for interpreting grammatical relations. Urak Lawoi’ and Singapore Bazaar Malay share these characteristics, as do Nonthaburi Malay (Tadmor 1995), Moken and Moklen (Larish 2005).
Pattani Malay as spoken in southern Thailand and the Kelantan dialect both appear to retain an alternation between an A-oriented transitive construction and a P-oriented transitive construction (two-way symmetrical voice), with special marking of the A argument in the P-oriented construction (Tadmor 1995: 249–250; Mahmood 1994: 201–207). The P-oriented construction in these varieties differs somewhat from the pv construction in Standard Malay/Indonesian. The pv prefix di‑ is not used to mark the verb; instead the morpheme (a)ɲo precedes the verb, while another particle, di/d or kɔ/kə/k precedes the A argument.7 This construction shows some formal similarities to analytic pv constructions found in some languages of Borneo (e.g. Land Dayak languages, see § 55.2, and North Sarawak languages, see § 55.3).
Negeri Sembilan Malay, on the other hand, shows morphological marking of pv with di‑ and optional prepositional marking for non-pivot A arguments (Hendon 1966: 68–69), and thus is similar to Standard Malay/Indonesian and languages of Java and Madura described above.
54.1 Language Contact, Restructuring, and the Loss of Applicatives
We know that restructuring of Chamic languages happened in mainland Southeast Asia due to contact with Austroasiatic languages. This restructuring is characterized by phonological changes, i.e., reduction of multisyllabic words and development of final word stress, morphological changes, i.e., reduction of affixation and shift toward analytic or isolating word structure, and syntactic changes, i.e., development of more fixed word order, loss of case distinctions, and use of periphrastic constructions to express meanings similar to morphologically marked causatives and ACs in other West Nusantara languages (Thurgood 1999, 2010; Brunelle 2020).
An example of a periphrastic construction with a benefactive meaning is given in (103). Here, the particle brei, which elsewhere may function as a full lexical verb meaning ‘give’, precedes the beneficiary, kơ ñu ‘for him’, which is always expressed as prepositional phrase in this type of construction. This construction is not an applicative as defined in this study because brei is an independent morpheme and its distribution is not closely tied to the verb or verbal complex; here the noun expressing the P argument, ƀrăm ‘arrow’, and its modifiers intervene between the closest verb and brei. Instead, this type of clause may be classified as a serial verb construction (see § 35).
(103) Bih, Periphrastic benefactive construction with ‘give’
Thô
gơ
magĭr
ngă
ana
năn
ngă
leh
ngă
ana
rĭ
ƀrăm,
dua
tlâo
urăt,
brei
kơ
ñu.
T.
3
pfx.try
make
crossbow
dist
make
pfv
make
crossbow
whittle
arrow
two
three
cl
ben
dat
3
‘Thô tried to make a crossbow and some arrows for him.’(Nguyen 2013: 90)
The pattern of restructuring observed for Chamic is also borne out for Moklen and Moken. Both lack applicatives, lack productive affixation, and show word final stress (Larish 2005). Moklen is also reported to make “heavy use of serial verb constructions” (Larish 2005: 527).
Malay varieties of Thailand and northern Peninsular Malaysia show similar changes to those found in Chamic at lower time depths. There is no productive affixation at all in Nonthaburi Malay. Pattani Malay retains a single affix, nominalizing /-ɛ/, and all other inherited affixes are found only in loanwords or remnant forms (including causative per‑, which is rare and has been reduced to initial gemination) (Tadmor 1995: 228–232). Kelantan Malay is similar to Pattani, except it retains limited use of ‑i as a causative or applicative suffix (Wu 2023). These varieties also form benefactive constructions by use of the verb meaning ‘give’, as shown in example (104) below. Urak Lawoi’, a Malay language spoken on islands in Southern Thailand, has no applicatives and shows several productive prefixes but no suffixes. Hogan (1988) writes that, “when [Urak Lawoi’] is compared with Bahasa Malaysia [Standard Malay], it is evident that it has a much smaller inventory of affixes, and that these are used much less frequently than the corresponding affixes in that language.” While Urak Lawoi’ has not developed word final stress, it shows signs of phonological reduction in non-final syllables, which include stressless “pre-syllables” and “minor syllables” that may bear stress but show a reduced number of phonemic contrasts in coda position compared to the final syllable (Hogan 1988: 11–12). A pattern of reduced affixation may also be found in Kedah Malay. Omar (1981: 7) writes, “in the Kedah dialect of North Malaya there is hardly any verbal suffix, while the nominal suffix ‑an has a very low frequency. The causative function of ‑kan is, in this dialect, fulfilled by the prefix pər‑ or a verbal phrase consisting of two verbs, while its benefactive function fulfilled by a prepositional phrase.”
(104) Nonthaburi Malay, Periphrastic benefactive construction with ‘give’
mɔʔ
bli
tpoŋ
bɣi
anɔʔ
makiʌŋ
mother
buy
snack
give
child
eat
‘The mother bought snacks for her children.’(Tadmor 1995: 261)
Finally, Singapore Bazaar Malay, used as a language of interethnic communication in Singapore, has also been shaped by contact with non-Austronesian languages, especially Hokkien (Sinitic). It employs only compounding and reduplication in word formation, and no use of affixation (Aye 2005: 62). Serial verb constructions with the verb kasi ‘give’ are used to express recipient and beneficiary participants (Aye 2005: 290–292).
In summary, the Austronesian languages of mainland Southeast Asia and northern Peninsular Malaysia show a strong tendency to lack applicatives, and a distinct typological profile compared to most other West Nusantara languages. This is the result of language contact with non-Austronesian languages, which caused changes in word structure, a shift away from morphological processes of affixation, and a shift towards analytic structures, including the use of serial verb constructions to express some meanings elsewhere associated with applicatives.
55 Borneo and the Southern Philippines
In this section, I present and discuss results of the typological survey for languages of Borneo and nearby areas of the Southern Philippines. Borneo is the largest major island in West Nusantara and the most linguistically diverse, with about 200 total languages spoken on the island. This section is organized by genetic grouping, and covers Sabahan languages (§ 55.1), Land Dayak languages (§ 55.2), North Sarawak and Melanau-Kajang (§ 55.3), Greater Barito languages, including Malagasy and Sama-Bajaw languages which are geographically dispersed outside of Borneo (§ 55.4), and languages spoken in Borneo belonging to other genetic groupings (§ 55.5). Map 8 gives an overview of the distribution of applicatives in the languages of Borneo and the Southern Philippines discussed in this section.



Map 8
Language overview for Borneo and the Southern Philippines
55.1 Sabahan Languages
As reflected in the their name, Sabahan languages are spoken across the Malaysian state of Sabah in northeastern Borneo, though some members of the group are also found in Brunei and parts of North Kalimantan Province adjacent to Sabah. As shown in Table 23, seven Sabahan languages are included in the sample.
Sabahan languages for the most part show productive Philippine-type voice systems (5 of 7 languages in the sample), and lack pivot-neutral applicatives (7 of 7). Sabahan languages are predominantly verb-initial (6 of 7) and typically exhibit systems of morphological alignment that mark the pivot grammatical relation distinctly from non-pivot core arguments (6 of 7 languages). They also tend to show complex systems of verbal morphology, as morphological marking for several tam categories appears on the verb (e.g. tense, volitionality, and non-indicative moods). In these respects Sabahan languages are more similar to Philippine and Formosan languages than other languages of Borneo. The common features of Sabahan languages noted here are broadly distributed, and are found across primary branches of Sabahan (i.e. Greater Dusunic [Kimaragang], Murutic [Timugon and Keningau Murut], and Paitanic [Tombonuo]).
Table 23
Typological survey results for Sabahan languages
|
Gen. grp. |
Name |
Applicatives? |
Voice |
Morph. align. |
Word order |
|
|---|---|---|---|---|---|---|
|
Pivot- |
Pivot- |
|||||
|
selecting |
neutral |
|||||
|
Sabahan |
Kimaragang |
Y |
N |
Philippine-type |
pivot-nonpivot |
vap/vpa |
|
Sabahan |
Keningau Murut |
Y |
N |
Philippine-type |
pivot-nonpivot |
vap |
|
Sabahan |
Timugon Murut |
Y |
N |
Philippine-type |
pivot-nonpivot |
vap/vpa |
|
Sabahan |
Tombonuo |
Y |
N |
Philippine-type |
pivot-nonpivot |
vap |
|
Sabahan |
Tatana |
Y |
N |
Philippine-type |
pivot-nonpivot |
vpa |
|
Sabahan |
Ida’an |
N |
N |
two-way |
pivot-nonpivot |
avp |
|
Sabahan |
Serundung Murut |
N |
Y |
two-way |
mixed-npiv.a |
vpa |
55.1.1 Pivot-Selecting Applicatives in Sabahan
Most Sabahan languages show between four and six contrastive voice categories, of which two or three are pivot-selecting applicatives (see discussion of voice and applicatives in Kimaragang in § 34). Tatana shows six contrastive voices, presented in example (105).8 Of these six constructions, four can be considered pivot-selecting applicatives, namely, theme voice (tv), in which the pivot is a theme or object undergoing transfer, instrument voice (iv), “referent voice” (rv), in which the pivot is a semantic beneficiary, recipient, or content item, and finally, locative voice (lv), in which the pivot indicates the time or location of the event expressed by the clause.9 Morphological marking for pivot-selecting applicatives is closely integrated with verbal morphology for tam in Sabahan languages. Table 24 presents a partial paradigm for verbal morphology in Tatana (Dillon 1994), which distinguishes volitive and non-volitive moods, and indicative and imperative moods, though not all tam categories may be marked in all of the six voices.10
Table 24
Partial paradigm for Tatana verbs
|
Voice category |
Volitive |
Non-volitive |
Imperative |
||
|---|---|---|---|---|---|
|
Non-past |
Past |
Non-past |
Past |
||
|
Actor voice (av tr) |
moN- |
noN- |
moko- |
noko- |
poN- |
|
Patient voice (pv) |
-on |
-in- -∅ |
mo- -∅ |
no- -∅ |
-o’ |
|
Theme voice (tv)† |
i- |
— |
moko- |
noko- |
— |
|
Instrument voice (iv)† |
poN- |
pinoN- |
— |
— |
— |
|
Referent voice (rv)† |
-an |
-in- -an |
mo- -an |
no- -an |
-i’ |
|
Locative voice (lv)† |
poN- -an |
pinoN- -an |
— |
— |
— |
† These voice categories constitute pivot-selecting applicative constructions.
(105) Tatana, Philippine-type voice alternations
a.
Mom-(p)opot
aku
do
daging
karabau.
av.tr.npst-chop
1sg.nom
dat
meat
buffalo
‘I am chopping buffalo meat.’ (av)(Dillon 1994: 22)
b.
Boli-in
ku
dudungu’
diti
buy-pv
1sg.gen
banana
this
‘I am buying these bananas.’ (pv)(Dillon 1994: 44)
c.
I-taak
ku
sita’
ku
dokou
cv-give
1sg.gen
shirt
1sg.gen
2sg.dat
‘I’ll give my shirt to you.’ (tv)(Dillon 1994: 47)
d.
Duit
pam-(b)ali
ku
do
kana’
money
iv-buy
1sg.gen
dat
fish
‘I buy fish with money.’ (iv)(Dillon 1994: 49)
e.
Bali-an
ku
okou
do
dudungu.
buy-rv
1sg.gen
2sg.nom
dat
banana
‘I am buying bananas for you.’ (rv)(Dillon 1994: 52)
f.
Kadai
diti
andang-andang
pam-(b)ali-an
ku
shop
this
rdp-usual
lv-buy-lv
1sg.gen
‘This shop is where I usually buy things.’ (lv)(Dillon 1994: 60)
55.1.2 Outliers in Sabahan
Two Sabahan languages in the sample, Ida’an and Serudung Murut, are outliers in the group. Both have lost Philippine-type voice, and thus lost pivot-selecting applicatives. Neither shows evidence of pivot-neutral applicatives.
The voice system for Ida’an has been reduced to av and pv for basic transitive constructions (Goudswaard 2005). The language shows other morphological and phonological characteristics unusual for Sabahan languages and more like that of the North Sarawak languages to the south, i.e., final stress, ablaut, and loss of all suffixation (Goudswaard 2005: 67–69). Ida’an also shows a shift away from verb-initial word order towards verb-medial word order (avp in av). Word order is an important indicator of grammatical relations, as only the pivot argument can precede the verbal predicate. Together, these are clues that Ida’an may have undergone a path of grammatical restructuring similar to that of many Borneo languages south of Sabah that completely lack applicatives, such as Land Dayak (§ 55.2) and North Sarawak languages (§ 55.3). It has also been suggested that Ida’an and Bonggi–both Northeast Sabahan languages—should be classified separately from the other (Southwest) Sabahan languages (Lobel 2013).
Serudung Murut is also analyzed as a two-way symmetrical voice system, with one A-oriented transitive construction (av) and one contrasting P-oriented transitive construction (pv) (Townsend 2017). Unlike other Sabahan languages, in Serudung Murut pivot arguments and non-pivot P arguments are marked alike morphologically (use of bare NP or nominative free pronoun), while non-pivot A arguments have distinctive marking (use of genitive linker nu or genitive enclitic pronoun), see example (106) below.
It is not entirely clear why Philippine-type voice was lost in Serudung Murut. However, it appears that extreme lexicalization of verbal suffixes has occurred, obscuring the distinction between patient voice and some former circumstantial voice (cv) category, i.e. a voice alternation in which an instrument, theme, or beneficiary is the pivot. Verb forms in undergoer-oriented transitive clauses in Serundung Murut may take either (i) suffixal marking ‑on/-oʔ or, (ii) suffixal marking ‑in/-iʔ in imperfective aspect and ‑an in perfective aspect (Townsend 2017: 27–29). The former is noted to occur with verbs that take patient-like P arguments, and the latter with verbs that take theme-like P arguments. Elsewhere in Sabahan, ‑on is the normal marker of patient voice, while ‑an/-in/-i(ʔ) mark a type of cv in various tam categories (as with Tatahan beneficiary-selecting rv discussed above).
Occurrences of ‑an/-in/-iʔ as verbal suffixes in Serundung Murut might thus be considered a type of degraded or non-productive pivot-selecting cv construction. However, the alternation between the two suffixal sets in the present-day language is completely lexically-conditioned; no single verb may take both sets, as shown in (106) for the verb taak ‘give’. The two suffixal sets therefore are not contrastive, and neither would mark a fully productive voice category on its own. For this reason, I consider Serudung Murut to represent a case of attrition of applicatives that can be attributed—at least in part—to lexicalization. Remnants of a former cv construction remain only as an apparent irregularity in the paradigm for morphological marking of pv.
(106) Serudung Murut, Voice alternations with ‘give’
a.
aku
an-(t)aak
lamun
sokou
1sg.nom
av-give
rice
2sg.obl
‘I give rice to you.’ (av)
b.
lamun
taak-in=ku
sokou
rice
give-pv=1sg.gen
2sg.obl
‘Rice is given to you by me.’ (pv)
c.
*
taak-on
*
ungrammatical,
not
a
possible
word
(Townsend 2017: 29–30)
55.2 Land Dayak Languages
On the other end of Borneo from Sabah, we find the Land Dayak languages, a group of about 15 languages spoken in southwestern Borneo, including the far western extent of Sarawak state and bordering parts of West Kalimantan Province, stretching south into inland West Kalimantan. There is little linguistic description available for most Land Dayak languages. Four languages are included in the survey: Benyadu’, Bakati’, Matéq and Ribun. Two additional languages for which short sketches are available—Bidayuh Serian and Biatah Bidayuh—will also be referenced (Omar 1983). Results of the survey for Land Dayak languages are shown in Table 25.
Table 25
Typological survey results for Land Dayak
|
Gen. grp. |
Name |
Applicatives? |
Voice |
Morph. align. |
Word order |
|
|---|---|---|---|---|---|---|
|
Pivot- |
Pivot- |
|||||
|
selecting |
neutral |
|||||
|
Land Dayak |
Benyadu’ |
N |
N |
two-way |
pivot-nonpivot |
avp |
|
Land Dayak |
Bakati’ |
N |
N |
two-way |
mixed-npiv.a |
avp |
|
Land Dayak |
Matéq |
N |
N |
two-way |
neutral |
avp |
|
Land Dayak |
Ribun |
N |
N |
two-way |
mixed-npiv.a |
avp |
All Land Dayak languages show no applicatives of any type (4 of 4 languages of the sample, plus Bidayuh languages). There is also no evidence that pivot-selecting applicatives can be reconstructed for Land Dayak, and no evidence that these languages ever had pivot-neutral applicatives.
Land Dayak languages generally have two-way symmetrical voice systems (4 of 4 languages of the sample). av constructions have unmarked avp word order in av constructions (4 of 4). In all four languages of the sample, word order is an important indicator not just of grammatical relations, but also the voice category of a transitive clause, as morphological marking for voice on the verb no longer clearly distinguishes av and pv, with the standard word order for pv being pav. Morphological marking for arguments distinguishing core grammatical relations also shows signs of reduction. Benyadu regularly makes use of pronominal sets distinguishing non-pivot A arguments from other core arguments. But in other Land Dayak languages this distinction is lost—as in Matéq, which has neutral alignment with all core argument marked the same—or almost entirely lost—being found only for third plural pronominal actors in Bakati, and optionally with second person singular pronominal actors in Ribun. In addition, Land Dayak languages show complete loss of suffixation.11 Here, as in Sumatra and mainland Southeast Asia, loss of suffixation appears to be one major factor associated with lack of applicatives.
Related to reduction of verbal morphology, the four Land Dayak languages of the sample show a shift towards analytic means of marking the distinction between av and pv. The verb in av clauses is marked with the prefix N‑ in Land Dayak, as is common in West Nusantara. Unusual however, is the marking of pv clauses, in which the verb also may bear N‑. This is shown in (107) below from Matéq. Here the morpheme marking pv is not a verbal affix, but a particle ni that precedes both the A argument (underlined), if expressed, and the verb bearing N‑. In Matéq both analytic pv and morphologically marked pv are available, but the latter only with certain verbs. Similar analytic pv constructions are found in Ribun, Benyadu, and Bakati’ (Sommerlot 2020), as well as Kendayan (Salako), a Malayic language spoken nearby in West Kalimantan and western Sarawak (Adelaar 2002). The particle marking pv in Ribun and Salako can be omitted if clausal word order is pav, showing that word order is a primary indicator of pv in these languages.
(107) Matéq, Analytic pv
a.
pingàt
aiq
yoh
ni
koq
moruh
plate
that
prt
pv
1sg
av.smash
‘I smashed the plate’
b.
ni
ular
aiq
degeq
nyora
ruba
turuaq=ng
pv
snake
that
constantly
av.attack
hole
dibbling.stick=3
‘the snake kept on attacking their dibbling holes’(Connell 2013: 113)
In summary, Land Dayak languages show a shift away from morphological marking for voice and grammatical relations and towards analytic means of indicating the same, especially in pv constructions. This trend and the loss of suffixation probably account for the loss of pivot-selecting applicatives and lack of pivot-neutral applicatives in Land Dayak.
55.3 North Sarawak and Melanau-Kajang Languages
This section covers the North Sarawak group and the Melanau-Kajang languages, which are spoken across north central Borneo to the south of Sabah. Melanau-Kajang languages are concentrated in central Sarawak state, while North Sarawak languages are spoken across a broad area covering northern and central Sarawak state, North Kalimantan and East Kalimantan. North Sarawak as defined in the study includes 41 total languages, with eight selected for the sample. Melanau-Kajang includes 12 total languages, of which just one (Central Melanau, Mukah dialect) was able to be included in the sample.
Many languages of these two groups lack adequate linguistic description and even basic documentation. Making matters more complicated, a number of subbranches of North Sarawak have a disputed classification, e.g. Kayanic languages, Kenyah languages, and the Punan-Mueller Schwaner languages (Eberhard, Simons & Fennig 2021, cf. Hammarström et al. 2022). On top of this, even languages within the same subbranch of North Sarawak are known to show important differences in their verbal systems (see e.g. B. Clayre 2005 on voice in the Dayic subbranch of North Sarawak). For this reason, in this section I will make mention of a number of languages not included in the survey sample for which some sketch material and limited description of verbal constructions are available. These are (i) for Kayanic languages, Segai (Soriente 2013), (ii) for Kenyah languages, the Lepo Ké dialect of Mainstream Kenyah, the Lebu’ Kulit dialect of Wahau Kenyah, and the Penan Benalui dialect of Western Penan (B. Clayre 1996; Omar 1983; Soriente 2013); and (iii) for Dayic languages, Sa’ban (B. Clayre 2005). Unfortunately, there are no grammatical descriptions for Punan-Mueller-Schwaner languages nor for Melanau-Kajang languages beside Central Melanau; these remain unrepresented, or nearly so, in the sample.
Table 26
Typological survey results for North Sarawak and Melanau-Kajang
|
Gen. grp. |
Name |
Applicatives? |
Voice |
Morph. align. |
Word order |
|
|---|---|---|---|---|---|---|
|
Pivot- |
Pivot- |
|||||
|
selecting |
neutral |
|||||
|
N. Sarawak |
Kelabit |
Y |
N |
Philippine-type |
neutral |
vpa |
|
N. Sarawak |
Lun Bawang |
Y |
N |
Philippine-type |
pivot-nonpivot |
avp/vpa |
|
N. Sarawak |
West Berawan |
N |
N |
two-way |
mixed-npiv.a |
avp |
|
N. Sarawak |
Belait |
N |
N |
two-way |
mixed-npiv.a |
avp/vpa |
|
N. Sarawak |
Baram Kayan |
N |
N |
two-way |
pivot-nonpivot |
avp |
|
N. Sarawak |
Eastern Penan |
N |
N |
two-way |
mixed-npiv.a |
avp |
|
N. Sarawak |
Punan Tubuh |
N |
N |
two-way |
mixed-npiv.a |
avp |
|
Melanau-Kajang |
Central Melanau |
N |
N |
two-way |
mixed-npiv.a |
avp |
Results of the survey are shown in Table 26. The large majority of languages of North Sarawak (6 of 8 languages in the sample) lack applicatives of any type, as does Central Melanau (1 of 1).12 Other Kenyah languages and the Dayic language Sa’ban also appear to lack applicatives (B. Clayre 2005; Soriente 2013). Within North Sarawak, just two languages (2 of 8) show applicatives; both Kelabit and Lun Bawang (Lundayeh) exhibit a pivot-selecting instrumental applicative (iv).
Of the North Sarawak languages lacking applicatives in the sample, all six have two-way symmetrical voice systems, as does Central Melanau. This appears to hold for Sa’ban (Dayic), Segai (Kayanic), and Western Penan (Kenyah). However, B. Clayre (1996: 74) reports that the Lepo Ké dialect of Mainstream Kenyah does not have an undergoer-oriented construction, which indicates that its voice system is asymmetrical. On the other hand, Kelabit (Philippine-type), shows av, pv, and iv voice categories (Hemmings 2016: 201–204), and Lun Bawang (marginal Philippine-type), shows av, pv, and an archaic iv category that is rare in current usage (Mortensen 2021: 115); for more detail see § 55.3.1 below.
Most languages of North Sarawak show a system of morphological alignment in which the non-pivot A argument has distinctive marking compared to other core arguments, which are marked alike (5 of 8). This is also found in Central Melanau. The distinctive form for non-pivot A is generally limited to pronominal arguments, and often found only in certain number and person categories, as in Punan Tubu, where there are distinct pronominal forms for first and second person singular non-pivot A arguments, and third singular non-pivot A can be expressed either with a distinct form or with the free pronoun form used for all other core arguments. In Sa’ban, distinctive forms of the pronouns for non-pivot A have been lost (neutral alignment), these being found only in fossilized remnants (B. Clayre 1996: 57). Two languages of North Sarawak (2 of 8; Lun Bawang, Baram Kayan) show one distinct marking for pivot arguments, and other types of marking for non-pivot core arguments (pivot-non-pivot). It looks like Kelabit formerly had such a system, but it has been collapsed such that core arguments have distinct marking distinguishing them from obliques, but are now not otherwise distinguished for grammatical relation (neutral alignment, Hemmings 2016: 331–332).
In terms of word order, most languages of North Sarawak show a preference for avp unmarked word order in A-oriented constructions (verb-medial; 5 of 8), which also holds for Central Melanau. A number of these languages also allow vpa order in av, but it is not always reported whether or not this is possible. In Belait and Lundayeh, word order is reported to be split between avp and vpa in av, with no indication of preference. In Kelabit, vpa is the most natural word order for av (verb-initial), but vap and avp are also possible. Thus, it appears that most North Sarawak languages show a shift away from preference for verb-initial word order as is typical in Philippine-type languages and most common in Sabahan languages. Strict ordering of verb + non-pivot core argument is common here, with this ordering being important for signaling grammatical relations, especially in av. For example, in West Berawan, word order is more strict in av than in pv; in av only avp is possible, while in pv, both pva and vap are found (B. Clayre 1997: 234).
Regarding verbal morphology and general morphological complexity, Lun Bawang and Kelabit are the most conservative. Only these two languages retain productive use of verbal suffixes, including pv ‑en and imperative suffixes, though the latter are reportedly rarely found in Lun Bawang (Hemmings 2016; Mortensen 2021). For all other languages of North Sarawak there is no evidence of productive suffixation, verbal or otherwise. In Central Melanau, there is one reported suffix, ‑ai, but it is found only on a very small number of directional adverbs (I. Clayre 1972: 210–211). Thus, as in languages of mainland Southeast Asia and Land Dayak languages, loss of suffixation is associated with complete lack of applicatives in North Sarawak and Melanau-Kajang.
A number of North Sarawak languages are reported to make use of an analytic or periphrastic pv construction, most commonly marked with the verb ‘do/make’. This is reported in Lun Bawang, Sa’ban, and Dayic varieties of East Kalimantan, as well, as Eastern Penan (Kenyah), Central Berawan and some Kayan varieties (see Mortensen 2021; B. Clayre 1996: 75–81, 2005).13 An example of this is given in (108) from Sa’ban, where this construction is replacing morphologically marked pv (B. Clayre 2005: 30). This construction can be considered a type of serial verb construction (see § 35), but not an applicative, as it does not select a peripheral semantic role as a core argument. It appears that no construction of this type is found in Central Melanau.
(108) Sa’ban, Periphrastic pv with ‘make’
Ayeu
noknai
an
ieh
m-paeng.
tree
this
make
3sg
av-cut.down
‘He will cut down this tree.’(B. Clayre 1996: 78)
In summary, Central Melanau and most languages of the North Sarawak group show loss of Philippine-type voice (pivot-selecting applicatives) and lack of pivot-neutral applicatives. Languages in this area that lack applicatives altogether also lack suffixation, and tend to show reduced verbal morphology. In some of these languages, we also see a shift towards analytic structures with word order becoming important for signaling grammatical relations and increasing use of periphrastic pv constructions, which represent a type of serial verb construction.
55.3.1 Outliers in North Sarawak
The only North Sarawak languages that retain Philippine-type voice alternations are Lun Bawang and Kelabit. Geographically, the areas where Lun Bawang and Kelabit are traditionally spoken stand in the northeast corner of Sarawak state, with Lun Bawang to the north of Kelabit.
Both Kelabit and Lun Bawang show an instrument-selecting voice construction (iv) that contrasts with av and pv (Hemmings 2016; Mortensen 2021), as in example (109). In this construction, shown in (109c), the instrument is always the clausal pivot (bolded); thus, this construction constitutes a pivot-selecting applicative. The iv construction in Lun Bawang is reportedly rarely used in spontaneous, spoken language, but it is still found in some written documents and “many speakers are aware of it and can produce it and judge its usage upon request” (Mortensen 2021: 115).
(109) Kelabit, Transitive voice alternations
a.
La’ih
sineh
ne-nekul
nubaq
nedih
ngen
seduk.
man
dist
pfv-av.spoon.up
rice
3sg.poss
with
spoon
‘That man spooned up his rice with a spoon.’ (av)
b.
Sikul
lai’h
sineh
nubaq
nedih
ngen
seduk.
pv.pfv.spoon.up
man
dist
rice
3sg.poss
with
spoon
‘That man ate his rice with a spoon.’ (pv)
c.
Seduk
pe-nekul
la’ih
sineh
nubaq
nedih.
spoon
iv-spoon.up
man
dist
rice
3sg.poss
‘That man used a spoon to spoon up his rice.’ (iv)(Hemmings 2016: 303)
Neither Lun Bawang nor Kelabit show a productive lv construction. In Kelabit, however, the suffix ‑an is retained in fossilized locative nominalizations, and a remnant of lv is found with a single verb, tu’an ‘to do/put, lv’ (Hemmings 2016: 145–146). In this construction, shown in (110), the phrase expressing the location of the event is the clausal pivot, as is indicated by its access to relativization in (110b).
(110) Kelabit, Remnant locative voice construction
a.
Lidung
tu’an
neh
babeh
nedih.
corner
do/put.lv
3sg.core.ii
bag
3sg.poss
‘He put his bag in the corner.’ (lv)(Hemmings 2016: 145)
b.
Seni’er
kuh
lidung
[suk
tu’an
neh
babeh
nedih].
pv.pfv.see
1sg.core.ii
corner
[rel
put.lv
3sg.core.ii
bag
3sg.poss]
‘I saw the corner [where he put his bag].’ (lv, relative clause)(Hemmings 2016: 218)
Kelabit and Lun Bawang belong to the Dayic primary branch of North Sarawak languages. Other Dayic languages, including Sa’ban, Lengilu’, and the Kerayan dialects, whether spoken to the south of Kelabit in Sarawak or to the east in North Kalimantan, have lost iv and now show a two-way voice system (B. Clayre 2005), which is also the norm for the rest of the North Sarawak group. Thus, it appears that North Sarawak languages show decay of the Philippine-type voice system, but this process is progressing slower in Kelabit and Lun Bawang than the rest of the group, as Kelabit and Lun Bawang have lost lv, while retaining iv, at least to some extent. This retention of iv should be viewed in light of a general pattern of linguistic conservatism for Lun Bawang and Kelabit, which is likely due to the two languages’ geographic position in the mountainous highlands, affording a degree of separation from other North Sarawak languages, including other Dayic varieties as well as Kenyah and Kayan languages present in the region (see Mortensen 2021: 232–256 on contact and dialect-mixing leading to spread of innovative features in Dayic outside of Lun Bawang and Kelabit).
55.4 Greater Barito Languages
Greater Barito is a linkage representing 35 languages that originated in the Barito River basin encompassing much of southwestern Borneo (Smith 2018). About 14 of these languages are spoken today near their homeland, in the Indonesian provinces of Central Kalimantan and East Kalimantan, while the others show an extremely broad geographic dispersion. Eight Greater Barito languages are included in the sample, as shown in Table 27. In the survey results, Greater Barito languages fall into three types. These are (i) the Malagasy languages of Madagascar in eastern Africa, which show Philippine-type voice and thus pivot-selecting applicatives, (ii) the Sama-Bajaw languages found in the Philippines and across West Nusantara, which show a combination of pivot-selecting and pivot-neutral applicatives, and (iii) languages spoken in the Barito River basin, which show no applicatives of any type.
Table 27
Typological survey results for Greater Barito languages
|
Gen. grp. |
Name |
Applicatives? |
Voice |
Morph. align. |
Word order |
|
|---|---|---|---|---|---|---|
|
Pivot- |
Pivot- |
|||||
|
selecting |
neutral |
|||||
|
Gr. Barito |
Merina Malagasy |
Y |
N |
Philippine-type |
pivot-nonpivot |
vpa |
|
Gr. Barito |
Southern Sama |
Y |
N |
Philippine-type |
mixed-npiv.a |
vpa |
|
Gr. Barito |
Central Sama |
Y |
Y |
marg. Philipp.-typ |
mixed:other |
vpa |
|
Gr. Barito |
Yakan |
Y |
Y |
marg. Philipp.-typ |
pivot-nonpivot |
vap/vpa |
|
Gr. Barito |
West Coast Bajau |
N |
Y |
two-way |
pivot-nonpivot |
avp |
|
Gr. Barito |
Ngaju |
N |
N |
two-way |
pivot-nonpivot |
avp |
|
Gr. Barito |
Ma’anyan |
N |
N |
two-way |
mixed-npiv.a |
avp |
|
Gr. Barito |
Paku |
N |
N |
two-way |
mixed-npiv.a |
avp |
55.4.1 Malagasy
The Malagasy languages represent a group of closely-related languages spoken on the island of Madagascar, some 3,800 miles from West Nusantara, off the eastern coast of Africa. All belong to Malagasic, a lower-level subgroup of the Greater Barito linkage, which falls under the same branch as Paku and Ma’anyan, which are spoken in the Barito River basin. Only one Malagasy language, Merina Malagasy, is represented in the sample, as it has the most available resources due to its status as the most prestigious variety. While the Malagasy languages are listed under 12 separate iso-639-3 codes, they are by all accounts very similar to one another, such that they are commonly referred to as “dialects,” and Rasoloson & Rubino (2005) state that they are “often so closely related to one another that a clear group classification is uncertain” (456). In some cases, the identification of the variety represented in available source material is not straightforward. For the purposes of this survey, most Malagasy languages would be quite similar in profile to Merina. Map 9 shows the location for Merina Malagasy.



Compared with other West Nusantara languages, Merina, like other Malagasy languages, is relatively syntactically conservative, while showing phonological and lexical innovations due to contact with Bantu languages in Coastal East Africa (Adelaar 2012). Merina retains a Philippine-type voice system with five contrastive voice categories (Pearson 2001). Three of these represent pivot-selecting applicatives: theme (or conveyance) voice (tv), dative voice (dv) and circumstantial voice (cv). The language is verb-initial, with the pivot in final position in unmarked word order. In Merina, only pronominal forms are distinguished morphologically for core grammatical relations. The pivot relation is marked with the nominative set of pronouns, while other non-pivot core arguments are marked with dative or genitive pronouns (pivot-non-pivot morphological alignment). With respect to these typological characteristics, Malagasy is quite similar to most Sabahan languages and quite different from other Greater Barito languages.
55.4.2 Sama-Bajaw Languages
Sama-Bajaw languages also show geographic dispersal outside of Borneo, though most are still spoken inside West Nusantara and nearby in the southern Philippines. In total there are nine Sama-Bajaw languages, which together constitute one exclusive subgroup. Four of these are represented in the sample, but in order to capture patterns in the diverse typological features of this group, I will also refer to three others in the discussion here: Pangutaran Sama, Sama Balangingi’, and Indonesian Bajau (Walton 1986; Gault 1999; Donohue 1996).
Sama-Bajaw languages represent a cline of typological characteristics showing gradual decline of Philippine-type voice (and thus pivot-selecting applicatives) and development of pivot-neutral applicatives. The voice and applicative systems of these languages fall into three types, as represented in Table 28. One set of languages (Southern Sama, Sama Balangingi’) has productive Philippine-type systems with five voice categories (av, pv, iv, bv, lv). A second set (Central Sama, Yakan, Panguturan Sama) shows marginal Philippine-type systems with reduced productivity or restricted distribution of lv and iv, and no use or only minimal use of bv. This set also has productive pivot-neutral applicatives. The third set (West Coast Bajau, Indonesian Bajau) shows two-way symmetrical systems and productive pivot-neutral applicatives only.
Table 28
Applicatives in Sama-Bajaw languages
|
Pivot-selecting only |
Both types* |
Pivot-neutral only |
|---|---|---|
|
Southern Sama |
Central Sama |
West Coast Bajau |
|
Sama Balangingi’ |
Yakan |
Indonesian Bajau |
|
Pangutaran Sama |
* These languages have marginal Philippine-type pivot-selecting constructions and productive pivot-neutral applicative constructions.
Central Sama will be used to illustrate Sama-Bajaw languages with marginal Philippine-type alternations and pivot-neutral applicatives. The voice systems of Yakan and Pangutaran Sama, are in large part very similar to Central Sama.
Central Sama has four transitive voice alternations, which I will call av, pv, iv, and lv. However, stems marked for the latter two voices—the pivot-selecting applicatives—have a different distribution that those marked for the former two. iv and lv stems appear mostly in subordinate clauses, often appear in coordination with nouns and as the object of a preposition, and may be lexicalized as locative or instrumental nominalizations. At the same time, stems marked with the iv prefix paN‑ and lv suffix ‑an1 show evidence of status as verbs, including co-occurrence with aspectual, modal, and imperative morphology (James 2017: 59–67). For these reasons, James considers such constructions to be voice alternations but calls them ‘minor voices’, distinguishing them from the much more common and broadly distributed av and pv constructions, as well as a separate passive construction in Central Sama.
Central Sama also has an applicative suffix ‑an2, which marks the verb when a beneficiary, recipient, goal, stimulus, or addressee is selected as core argument. In example sentences, we see that this suffix co-occurs with av, marked with aN‑ on the verb as in (111b), pv, which has no overt voice morphology and shows obligatory actor indexing on the verb, as in (111c), and the passive construction, marked with ‑in‑ without actor indexing, as in (111d).
(111) Central Sama, Applicative ‑an
a.
Am-(b)uwan
akū
badju’
ma
abagay=ku.
av-give
1sg.i
shirt
to
friend=1sg.ii
‘I will give a shirt to my friend.’ (av)
b.
Am-(b)uwan-an
akū
bagay=ku
badju’
av-give-appl
1sg.i
friend=1sg.ii
shirt
‘I will give a shirt to my friend.’ (av + appl)
c.
B’lla-han=ta
ka
buwas
itu.
pv.cook-appl=1sg.ii
2sg.i
rice
this
‘I will cook you this rice.’ (pv + appl)
d.
B⟨in⟩uwan-an
kamī
kaldero.
⟨pass⟩give-appl
1pl.i
pot
‘We were given a cooking pot.’ (Passive + appl)
On the other hand, this applicative ‑an2, which selects one of the aforementioned semantic roles as a core argument, never co-occurs with the ‘minor voices’, iv or lv. Conversely, it is clear that iv and lv cannot co-occur with av, pv, or passive constructions. Instrument or theme-selecting paN‑ and location-selecting ‑an1 never co-occur with av aN‑ or passive ‑in‑. Also, while pv is not marked with distinct verbal morphology (i.e., it is zero-marked), it has obligatory actor-indexing on the verb, while iv and lv do not, as shown in (112) and (113) below. Thus, an analysis of paN‑ and ‑an1 as applicatives that combine with (only) pv is not supported. Central Sama therefore represents a language in transition from a robust Philippine-type voice system to a two-way symmetrical voice system plus pivot-neutral ACs.
(112) Central Sama, iv Construction
Sīn,
limangibu
bay
pam-(b)uwan
aku.
money
5,000
pst
iv-give
1sg.iii
‘Money, five thousand (pesos) was given me.’ (iv)James 2017: 63
(113) Central Sama, lv Construction
Waí
al’ssu’
kaldero,
mbal
na
ta-pam-(b)’lla-han
daing.
pfr
cracked
pot,
neg
now
abil-punct-cook-lv
fish
‘The pot has cracked, it can no longer be used for boiling fish’ (lit. place for boiling fish). (lv)(James 2017: 61)
If we compare languages like Central Sama with languages with more productive Philippine-type systems in the group, i.e. Sama Balangingi’ and Southern Sama, it becomes clear that pivot-neutral applicatives marked with ‑an in Sama-Bajaw are in complementary distribution with the voice category that selects a beneficiary pivot (bv) in Sama-Bajaw. This voice construction is also marked with ‑an in indicative mood. Thus, the development of pivot-neutral applicatives in Sama-Bajaw is directly related to the breakdown of the Philippine-type voice system. This is discussed in further detail in Chapter 6, where I argue that many pivot-neutral applicatives of West Nusantara showing selection of beneficiary, theme, and/or instrument applied phrases are derived from earlier Philippine-type circumstantial voice (cv) alternations.
55.4.3 Languages of the Barito River Basin
Besides Malagasy and the Sama-Bajaw languages, the remaining Greater Barito languages are still spoken near the Barito River basin in southeastern Borneo. In total, there are 14 such languages, but only 3 are represented in the sample, due to the general paucity of descriptive materials for Bornean languages, which is especially sparse for inland parts of the Malaysian state of Sarawak, and the Indonesian provinces of West, Central, and East Kalimantan.
In the sample, the Greater Barito languages of the Barito River area show similar typological characteristics to other Borneo languages spoken south of Sabah. All have no applicatives of any type (3 of 3). They show two-way symmetrical voice systems (3 of 3), systems of morphological alignment with distinctive marking for the non-pivot A argument (3 of 3), and predominant avp word order in av (3 of 3), with increasing reliance on word order for indicating grammatical relations (see Diedrich 2018: 182; Hardeland 1858: 161; Gudai 1985: 147–148).
Greater Barito languages of southeastern Borneo also show reduced suffixation, though not as completely as Land Dayak and most North Sarawak languages. Ma’anyan shows a single suffix (i.e. nominalizing ‑an), and two productive circumfixes (i.e. adversative kV- -an and nominalizing pVN- -an), but 11 productive prefixes (Gudai 1985). Paku also shows a single suffix, again nominalizing ‑an, but 16 total prefixes (Diedrich 2018). In Ngaju the suffix ‑an is reported to be rare, found only on a small number of nominalizations, and a few causative verbs also bearing the prefix ma‑ (Hardeland 1858: 64–66). Including ma‑, Ngaju has 16 prefixes.
Looking at the broader picture, the presence of Philippine-type voice in Malagasy and some Sama-Bajaw languages shows that Philippine-type voice was likely present in early stages for Greater Barito languages. In all likelihood, Philippine-type voice was only lost in languages of the Barito River basin after ancestral Malagasy speakers left Borneo and started their long and storied migration to Madagascar, sometime after ad 400 and most likely in the 7th century ad (Adelaar 2009). This loss probably also post-dates the departure of ancestral Sama-Bajaw speakers from the Barito River basin, sometime between ad 670–800 (Blust 2005). So while Greater Barito languages of the Barito River basin area show some similarities with Land Dayak and North Sarawak languages, and may have undertaken a similar path of grammatical restructuring resulting in the complete loss of applicatives, it is likely that these changes happened much later for Greater Barito languages like Paku, Ma’anyan, and Ngaju than for Land Dayak languages, and probably most North Sarawak languages as well.
55.5 Other Languages Spoken in Borneo
In addition to the indigenous languages of Borneo discussed already in this section, we also find languages spoken in Borneo which genetically belong to the Malayic and South Sulawesi genetic groups. Some of these kept what appear to be inherited systems of pivot-neutral applicatives after arrival at their present-day location in Borneo, including Brunei Malay, Kendayan (Salako), and Embaloh (see also discussion of Malayic languages of Sumatra in § 51 and discussion of South Sulawesi languages in § 56.1).
Some Malayic languages spoken in Borneo, however, appear to have lost inherited pivot-neutral applicatives and taken on some resemblance to other Borneo languages spoken south of Sabah in use of analytic constructions. These include Mualang, which is included in the sample, and Iban and the Belangin variety of Kendayan, which are not included in the sample, but are discussed here on the basis of available descriptive material (Omar 1981; Adelaar 2006).
Mualang and Belangin, are both spoken in West Kalimantan province, in the area that is also home to the Land Dayak languages. Both show no applicatives of any type, and no suffixation, showing loss of the applicative suffixes *-iʔ and *-an in Proto Malayic (Tjia 2007; Adelaar 2006).
While Mualang has lost applicative marking, it still shows alternations in clausal structure that I will refer to as “unmarked applicative analog” constructions. Examples are given in (114) and (115). For comparison, Indonesian ACs with suffixal AM marking are given in (116) and (117).
(114) Mualang, Unmarked benefactive construction
a.
Inay
N-beli
kayin
baju
ka
ia.
mother
av-buy
clothes
shirt
to
3sg
‘Mother bought clothes for her.’ (Prototypical av)
b.
Inay
N-beli
ia
ka
kain
baju.
3pl
av-buy
3sg
to
clothes
shirt
‘Mother bought her clothes.’ (Beneficiary-selecting construction)(Tjia 2007: 174)
(115) Mualang, Unmarked locative construction
a.
Ia
N-isi’
ay
ka
kuali.
3sg
av-content
water
to
cooking.pan
‘She is putting water into the pan.’ (Prototypical av)
b.
Ia
N-isi’
kuali
ka
ay.
3sg
av-content
cooking.pan
to
water.
‘She is filling the pan with water.’ (Goal-selecting construction)(Tjia 2007: 175)
(116) Indonesian, Beneficiary-selecting applicative
a.
John
Dul
mem-beli
buku
itu
untuk
Mary
Yem.
J.
D.
av-buy
book
that
for
M.
Y.
‘John bought that book for Mary Yem.’ (BC in av)
b.
John
Dul
mem-beli-kan
Mary
Yem
buku
itu.
J.
D.
av-buy-ben.appl
M.
Y.
buku
itu
‘John Dul bought Mary Yem a book.’ (AC in av)(Kaswanti Purwo 1995: 79)
(117) Indonesian, Goal-selecting applicative
a.
Mary
Yem
men-(t)uang(‑kan)
air
ke
ember.
M.
Y.
av-pour(‑thm.appl)
water
to
bucket
Y.
‘Mary Yem poured water into the bucket.’ (BC in av)
b.
Mary
Yem
men-(t)uang-i
ember
dengan
air.
M.
Y.
av-pour-loc.appl
bucket
with
water
‘Mary Yem filled the bucket with water.’ (AC in av)(Kaswanti Purwo 1995: 80)
While many languages allow for unmarked alternations of this kind with a limited number of semantically ditransitive verbs, in Mualang, we see a quite large of range of possible lexical verbs and verbal meanings used with such constructions. These examples show alternation in the realization of arguments for the semantic patient/theme and beneficiary/locative role. The existence of these alternations suggest that the morphological marking for applicative alternations in Mualang has been lost, but they remain as clausal constructions with distinctive argument structure, and these are still associated with the former meanings. I do not know of any other languages of West Nusantara that systematically exhibit this type of unmarked alternation.
In Belangin, the functions of ACs have been taken over, at least in part, by serial verb constructions (Adelaar 2006: 78). This is shown in example (118), where Belangin uses beri ‘give’ + minjam ‘borrow av’ to mean ‘lend to (s.o.)’. For this same meaning, the closely-related variety Salako, uses ng-inyapm-iʔ from the same root with the applicative suffix ‑iʔ.
(118) Belangin, Serial verb construction with ‘give’
Ba
beri
aku
m-(p)injam
sa-bantar
ba!
hort
give
1sg
av-borrow
one-moment
hort
‘Why don’t you lend it to me for a while!’ (about a flute)(Adelaar 2006: 78)
Iban, another Malayic language spoken in Central Sarawak, has also lost the original Proto Malayic applicative suffixes. According to Adelaar (2006), Iban subsequently developed a new causative/benefactive AM ‑-ka from the preposition ka, noting “that in older Iban sources ‑ka is still written as a separate word” (79). Besides ‑ka, Iban has no other suffixes but shows nine prefixes (Omar 1981).
While Adelaar (2006) connects the loss of suffixation in Iban and Belangin to an areal pattern also found in Land Dayak languages, he does not mention Mualang, nor consider the many North Sarawak languages which also lack all suffixation. As shown in the results presented in § 55.2-55.4, the areal pattern for reduced suffixation extends to a greater geographic area than previously recognized, and this pattern is broadly associated with complete lack of applicatives in Borneo south of Sabah. Possible explanations for this pattern are discussed in the next section.
55.6 Possible Explanations for the Lack of Applicatives in Borneo South of Sabah
Based on the results of the typological survey and other available descriptive material, among the languages of Borneo, loss of suffixation looks to be a major precipitating factor for the complete lack of applicatives, which is by far the predominant pattern in languages spoken south of Sabah. These languages mostly retain a two-way symmetrical voice alternation (certain Kenyah varieties reported to lack pv are exceptions). Reduction of verbal morphology, and a shift to reliance on word order for signaling grammatical relations, as well as a shift towards periphrastic or analytic signaling of pv are also observed, as is use of analytic structures with benefactive, locative, or causative meanings in some languages. This shift looks to be related to the loss of suffixation, including grammatical restructuring following the loss of morphological marking of pv with suffixal ‑en.
In mainland Southeast Asia and northern Peninsular Malaysia (as well as Singapore Bazaar Malay), loss of suffixation was triggered by language contact with non-Austronesian languages (Tadmor 1995; Thurgood 1999; Aye 2005). On the other hand, it is unknown why loss of suffixation occurred in so many Bornean languages, cutting across several genetic affiliations. Little is known of possible language contact for speakers of Malayo-Polynesian languages with non-Austronesian languages upon arrival in Borneo. The present-day pattern for lack of suffixation might be observed however, if some such contact was centered on the southwest part of Borneo, where Land Dayak languages are presently spoken, and these changes only later spread to North Sarawak, Melanau-Kajang and, finally, Greater Barito languages.
Contact-induced spread leading to loss of suffixation may have been chained, and might had have multiple centers, with possible differences in the precise path of historical change. In the case of Sa’ban (Dayic, North Sarawak), for example, it is argued that contact with Modang (Kayanic) set off the phonological changes that resulted in its highly reduced system of affixation (Blust 1999), in contrast to the relatively rich systems of Kelabit and Lun Bawang (also Dayic). Some Malayic languages whose speakers settled in southwestern Borneo also seem to have undergone loss of suffixes known to be present in Proto Malayic, e.g. Belangin, Mualang, and Iban, but others retain some inherited suffixes, e.g. Salako (Adelaar 2005b). It is unclear whether the nature or extent of the language contact can explain these differences, and this further complicates a proposed timeline for the spread of these changes. Nonetheless, it is clear that reduced suffixation is an areal pattern for Borneo south of Sabah, and that this pattern was influenced by complex histories of migration, contact, and genetic inheritance for Bornean languages and peoples. Still, the possibility that the decline of suffixation and subsequent changes in the verbal system were initially set off by contact with non-Austronesian speakers is an enticing prospect, albeit a speculative one at present.
56 Sulawesi
Sulawesi is the second largest island in West Nusantara after Borneo, and like Borneo also shows great linguistic diversity, with over 100 languages spoken on the island. Sulawesi is traditionally divided into microgroups, which I have generally adopted as genetic groupings for the purpose of the survey, though I have excluded three traditional microgroups that are thought to subgroup with Philippine languages (i.e. Gorontalo-Mongondow, Minahasan, and Sanggiric, see Blust 1991). In this section, I will consider separately languages of the South Sulawesi microgroup in § 56.1, and languages classified as Tomini-Tolitoli, Kaili-Pamona, Saluan-Banggai, Bungku-Tolaki, Muna-Buton, and Wotu-Wolio in § 56.2. This second set of genetic groupings is more varied in typological profile and have been proposed to belong to a genetic supergroup, Celebic (Mead 2003), that excludes the South Sulawesi languages. Map 10 shows languages spoken in Sulawesi in the sample. In addition to the languages discussed in this section, Indonesian Bajau is spoken in Sulawesi (see § 55.4.2).
56.1 South Sulawesi Languages
The South Sulawesi genetic group includes 30 total languages, which are primarily spoken on the west and southwestern side of Sulawesi. Two languages of this group are spoken in Borneo (i.e. Embaloh, Taman). In available documentation, these two appear to be quite similar to other South Sulawesi languages (Adelaar 1994). Seven South Sulawesi languages are included in the sample, as shown in Table 29. All of the South Sulawesi languages in the sample show pivot-neutral applicatives (7 of 7). They also share common typological features, especially in terms of voice system and morphological alignment.



Almost all South Sulawesi languages in the sample show what I have called a marginal two-way symmetrical voice system (6 of 7). These languages show an alternation between an A-oriented and a P-oriented construction, with the A-oriented construction showing reduced semantic transitivity (B. Friberg 1991: 105–112) or very limited distribution (Campbell 1989: 66). Here, reduced semantic transitivity refers to lower effectiveness or intensity in the acting of one entity on another. Constructions with reduced semantic transitivity may thus express non-telic or non-punctual events or states, lower affectedness of P, and/or lower individuation of P, among other properties (Hopper & Thompson 1980). Such A-oriented constructions are discussed in greater detail below (see also the discussion of Makasar and Bugis in § 73). In the sample, only Duri is considered a two-way symmetrical voice language (1 of 7). Valkama (1993: 69, 79–89) shows that the av construction in Duri does not have lower transitivity than the pv construction, and though it is less frequent in usage than the pv construction, av does not appear to be rare.14
Table 29
Typological survey results for South Sulawesi languages
|
Gen. grp. |
Name |
Applicatives? |
Voice |
Morph. align. |
Word order |
|
|---|---|---|---|---|---|---|
|
Pivot- |
Pivot- |
|||||
|
selecting |
neutral |
|||||
|
S. Sulawesi |
Embaloh |
N |
Y |
marg. two-way |
ergative |
vap |
|
S. Sulawesi |
Bugis |
N |
Y |
marg. two-way |
ergative |
avp/vpa |
|
S. Sulawesi |
Coastal Konjo |
N |
Y |
marg. two-way |
ergative |
vap |
|
S. Sulawesi |
Makasar |
N |
Y |
marg. two-way |
ergative |
vpa |
|
S. Sulawesi |
Bambam |
N |
Y |
marg. two-way |
ergative |
avp |
|
S. Sulawesi |
Seko Padang |
N |
Y |
marg. two-way |
ergative |
undet. |
|
S. Sulawesi |
Duri |
N |
Y |
two-way |
ergative |
avp |
All seven South Sulawesi languages of the sample are coded as showing ergative patterns of morphological alignment. This is because S in intransitive clauses and P in P-oriented transitive clauses pattern together in morphological encoding; both are typically indexed on the verb by means of a pronominal enclitic, while A in P-oriented clauses is typically indexed with a proclitic, as shown in (119a–b) from Makasar. In Seko Padang, two sets of proclitics are used instead of one enclitic set and one proclitic set (Payne & Laskowske 1997: 426–429). However, there are some complications for indexing of arguments in these languages that make them difficult to classify for type of morphological alignment.15
(119) Makasar, Voice alternations
a.
Tinroi
i
Ali
tinro=i
i
Ali
sleep=3
pn
A.
‘Ali is sleeping.’ (Intransitive)(Jukes 2020: 246)
b.
Kukanrei
untia
ku=kanre=i
unti=a
1=eat=3
banana=def
‘I eat the bananas.’ (Transitive, P-oriented)(Jukes 2020: 135)
c.
Angnganrea’
unti
aN(N)-kanre=a’
unti
biv-eat=1
banana
‘I eat bananas.’ (Semi-transitive, A-oriented)(Jukes 2020: 135)
South Sulawesi languages show some diversity of word order patterns. In three languages of the sample, verb-initial order is predominant, while in two languages (Duri and Bambam), pivot-initial order is predominant (avp in A-oriented constructions). In Bugis, word order is described as split between avp and vpa in the A-oriented construction. Word order is not discussed in available source material on Seko Padang.
As mentioned above, most South Sulawesi languages show an A-oriented construction with reduced semantic transitivity. An example is given in (119c) above from Makasar. Determining the syntactic transitivity of these constructions is not straightforward. While the P argument in such constructions typically is strictly indefinite, often it is overtly expressed, and it may even be obligatory to mention P overtly, as in Makasar and Bugis (Jukes 2020: 250–253; D. Laskowske 2016: 26). These constructions have been labeled in various ways by authors, but it is not clear whether they should be considered intransitive (e.g. “antipassive” as used by B. Friberg 1991), transitive (e.g. “actor focus” as used by Campbell 1989) or somewhere in the middle (e.g. “semi-transitive” as used by Jukes 2020).
It is interesting, then, that in languages of this type, pivot-neutral applicative suffixes typically may co-occur with both the more prototypically transitive P-oriented construction and the A-oriented construction that shows lower semantic transitivity. Some examples of ACs in A-oriented constructions are given below from Coastal Konjo in (120) and Embaloh in (121).
(120) Coastal Konjo, Locative applicative in A-oriented clause
a.
Ammalu’a
(tappere).
ang-halu’-a
tappere.
av-roll.up-1sg.abs
mat
‘I roll up a mat.’ (BC)
b.
Ammaliikia
palungang.
ang-halu’-i-a
palungang
av-roll.up-loc.appl-1sg.abs
pillow
‘I roll up a pillow (in something).’ (AC)(B. Friberg 1991: 115)
(121) Embaloh, Locative applicative in A-oriented clause
Urtkaʔ-ak
i-asan
marampas
tahuʔ
tau,
ingka,
i-asan
deʔi
man-jolo-i
laki
tau.
possibly-1sg.abs
erg-name
av.snatch
fiancé
person
said
erg-name
aforementioned
maN-take.away.place.next.to-loc.appl
husband
person
‘They might think that I’m after other people’s fiancés, she said, they might think that I tried to take away other women’s husbands’.(Adelaar 1995: 391)
In the Embaloh example above, the locative applicative suffix ‑i occurs on the verb man-jolo-i ‘to take away the place next to’ (cf. jolo ‘preceding, ahead of’). This verb also bears the maN‑ prefix associated with low referentiality of P or oblique-marking of the semantic undergoer, i.e. the entity that is understood to be acted upon in the event expressed.
Lastly, I will note that though these languages may show ergative morphological alignment, they are not necessarily syntactically ergative. While the ergative pattern for morphological alignment stems from the distribution of clitic pronominal forms, these clitic forms do not co-vary neatly with syntactic behavior. For Makasar, Jukes (2020) writes:
… the fact that in the majority of cases =abs corresponds to S and P, while erg= corresponds to A, suggests that clitics cross-reference core arguments, and that they do this according to an ergative-absolutive pattern. … However, when examining the behaviour of certain Makasar clauses it becomes clear (a) that it is not always easy to correlate =abs or erg= clitics with particular grammatical relations, and (b) that some arguments which are arguably core are not always cross-referenced. (331)
Other authors make similar observations for other South Sulawesi languages. Campbell (1989) writes: “Since, however, the ergative characteristics are limited to the pronominal system, pus [Bambam] is morphologically ergative in a very limited sense. Like many languages which have ergative morphology, pus does not have ergative syntax” (56). Of Coastal Konjo, Barbara Friberg (1991) writes “Only here [in clitic marking] is there an ergative system functioning; syntactically Konjo functions as an accusative system” (106–107). Thus, South Sulawesi languages may be morphologically ergative, but they do not necessarily provide evidence to support a correlation between non-accusative alignment and the presence of applicatives as found elsewhere (Peterson 2007).
56.2 Celebic Languages
The remaining languages of Sulawesi in the sample belong to six microgroups treated as genetic groupings for the purpose of the study: Tomini-Tolitoli, Kaili-Pamona, Saluan-Banggai, Bungku-Tolaki, Muna-Buton, and Wotu-Wolio. These microgroups have been proposed to belong to a supergroup, called Celebic, which includes 64 total languages, spoken across central, eastern, and southeastern Sulawesi. Of these, 20 languages are included in the sample, as shown in Table 30. Note that the genetic classification used is disputed for Totoli (representing the Tolitoli subgroup that may be non-Celebic, rather than subgroup with Tomini) and Behoa (representing the Badaic subgroup, which may belong with the Seko languages of South Sulawesi rather than Kaili-Pamona).
Almost all of the languages of the sample in these microgroups show pivot-neutral applicatives (29 of 30). Geographically, these languages are broadly distributed in Sulawesi. The presence of pivot-neutral applicatives is also observed to cut across other typological features of language in these microgroups, especially voice, morphological alignment and word order.
The languages of the sample in these affiliations are diverse in voice system. Most are two-way symmetrical systems (11 of 20), but asymmetrical voice systems (5 of 20)16 and Philippine-type systems (3 of 20) are also found. There is
Table 30
Typological survey results for other Sulawesi languages
|
Gen. grp. |
Name |
Applicatives? |
Voice |
Morph. align. |
Word order |
|
|---|---|---|---|---|---|---|
|
Pivot- |
Pivot- |
|||||
|
selecting |
neutral |
|||||
|
Tomini-Tolitoli |
Totoli* |
Y |
Y |
Philippine-type |
mixed-npiv.a |
avp/vpa |
|
Tomini-Tolitoli |
Dampelas |
N |
Y |
two-way |
mixed-npiv.a |
avp |
|
Tomini-Tolitoli |
Pendau |
N |
Y |
two-way |
mixed-npiv.a |
avp |
|
Tomini-Tolitoli |
Tajio |
N |
Y |
two-way |
mixed-npiv.a |
avp/vpa |
|
Kaili-Pamona |
Behoa* |
N |
Y |
two-way |
pivot-nonpivot |
avp |
|
Kaili-Pamona |
Ledo-Kaili |
N |
Y |
two-way |
mixed-npiv.a |
avp/vpa |
|
Kaili-Pamona |
Moma |
N |
Y |
two-way |
mixed-npiv.a |
avp |
|
Kaili-Pamona |
Uma |
N |
Y |
two-way |
ergative |
avp |
|
Kaili-Pamona |
Da’a Kaili |
N |
N |
two-way |
mixed-npiv.a |
avp |
|
Saluan-Banggai |
Balantak |
Y |
Y |
Philippine-type |
pivot-nonpivot |
avp |
|
Saluan-Banggai |
Bobongko |
Y |
Y |
marg. Philipp.-typ |
mixed-npiv.a |
undet. |
|
Bungku-Tolaki |
Mori Bawah |
N |
Y |
asymmetrical |
ergative |
vap/vpa |
|
Bungku-Tolaki |
Moronene |
N |
Y |
asymmetrical |
mixed:other |
vap/vpa |
|
Bungku-Tolaki |
Tolaki |
N |
Y |
asymmetrical |
split-S |
undet. |
|
Muna-Buton |
Busoa |
N |
Y |
asymmetrical |
accusative |
vap/vpa |
|
Muna-Buton |
Muna |
N |
Y |
asymmetrical |
accusative |
vap/vpa |
|
Muna-Buton |
Tukang Besi |
N |
Y |
two-way |
mixed:other |
vpa |
|
Wotu-Wolio |
Laiyolo |
N |
Y |
two-way |
accusative |
avp/vpa |
|
Wotu-Wolio |
Wolio |
N |
Y |
marg. two-way |
accusative |
vap |
|
Wotu-Wolio |
Wotu |
N |
Y |
two-way |
mixed:other |
avp |
* For these languages, the genetic classification listed is disputed.
also one marginal two-way system (Wolio retains an alternation between A-oriented and P-oriented clauses only in certain “participial” forms, see Anceaux 1952) and one marginal Philippine-type language (Bobongko, which appears to shows limited usage of lv and iv constructions, see § 56.3.3).
Languages of these six genetic groups have diverse systems of morphological alignment, with all seven coding categories for this feature represented, and pivot-neutral applicatives attested across all seven.
These languages also show diversity in word order patterns, with some languages preferring verb-initial orders (6 of 20), some favoring verb-medial orders (avp in A-oriented constructions, 8 of 20), and some showing split word order with flexibility for the pivot (avp/vpa word order in av, 4 of 20). For Bobongko and Tolaki, word order patterns could not be determined on the basis of available source material.
An unexpected finding in these data, is the fact that a few Philippine-type languages of Sulawesi also have a system of pivot-neutral applicatives which co-occur across different voice constructions, though not necessarily all such categories. This occurs in Balantak and Bobongko, two Saluan-Banggai languages of East Sulawesi province, and Totoli, a Tolitoli language of Central Sulawesi province. All three of these languages are located roughly in the northern third or so of Sulawesi, and are discussed in greater detail in the following subsection, along with Pendau, which shows some similarities to these three in its applicative system. I consider these to represent a transitional state between the two types of applicative systems.
The one language of Sulawesi in the sample that does not show applicatives is Da’a (Kaili-Pamona). Da’a lacks the ‑i locative AM (Martens 1988b: 193) found in other Kaili-Pamona languages. Also while Barr (1988a) reports limited use of ‑(C)aka as a causative marker, he does not mention or include data showing applicative functions. Martens (1990: 193) reconstructs Proto Pamona-Kaili (ppk) *-aka, “a transitivizer or causativizer or intensivizer”, with Da’a nan-taji-aka ‘to throw away’ as a reflex, and also tentatively reconstructs ppk *-uli’-ka ‘to tell, to inform’, an apparent addressee-selecting applicative verb, with Da’a nang-uli-ka as a reflex. It is possible that the applicative functions of the relevant suffixes have become unproductive in Da’a, but with only limited descriptive material available it is hard to state this definitively.
56.3 From Philippine-Type Voice to Pivot-Neutral Applicatives
Three languages of Sulawesi in the sample show Philippine-type voice alternations in addition to pivot-neutral applicatives: Balantak, Totoli, and Bobongko. Like the Sama-Bajaw languages discussed in § 55.4.2 above, these languages generally show reduction in the four-way voice system, and can be considered to show a transition between Philippine-type voice (pivot-selecting applicatives) and pivot-neutral applicatives. In addition, Pendau, a Tomini language of Central Sulawesi, shows irregularities in marking of voice and applicatives that suggest it has taken a path of development similar to Balantak, Totoli, and Bobongko. For this reason, it is also discussed in this section. Together, these four cases indicate that the transition between Philippine-type voice and pivot-neutral applicatives takes place via reduced productivity of the cv and/or lv voice categories, reinterpretation of certain affixal marking for voice (e.g. pog‑/paN‑, ‑an, and ‑i), and functional replacement of cv and/or lv with pivot-neutral ACs.
56.3.1 Balantak
Balantak is the most straightforward of these cases. Balantak has three basic transitive constructions, av, pv, and lv, but no circumstantial voice (cv) construction.17 It also has three applicative suffixes, which each may co-occur together with any of the three basic voices. The general applicative suffix ‑kon is shown in example (122) with the verb bisara ‘speak’. In clauses marked with ‑kon, one of a wide range of peripheral semantic roles is selected as a core argument. The applied phrase may be a semantic recipient, beneficiary, instrument, purpose, comitative, or content, among others (van den Berg & Busenitz 2012: 97–99). As shown in (122a), without applicative marking, bisara is intransitive. With applicative ‑kon, the verb selects a content applied phrase, and this may co-occur with all three transitive voice categories as shown in (122b–122d). Thus, Balantak has a productive Philippine-type system with a three-way distinction for voice, and shows pivot-neutral applicatives that are outside of the symmetrical voice paradigm. In Balantak, a number of functions of cv in other Philippine-type languages (selection of a beneficiary or instrument as a core argument) are filled instead by pivot-neutral applicatives.
(122) Balantak, Applicative ‑kon
a.
Kai
ba-bisara
sang-ilio.
1pl.excl
intr-speak
one-day
‘We talked for a whole day.’ (Intransitive)
b.
Kai
nim-bisara-kon
parakala
i-ya’a
1pl.excl
av.rls-speak-appl
issue
deic-dem
‘We talked about the issue.’ (av + appl)
c.
Parakala
men
bisara-kon-on-ta …
issue
rel
speak-appl-pv.irr-1pl
‘The issue that we will talk about …’ (pv + appl)
d.
Na
laigan-mo
ka-ni’i
a
bo
pim-bisara-kon-an-ta
parakala.
loc
house-pfv
deic-dem
art
for
ger-speak-appl-lv-1pl
issue
‘Let this be the house where we will talk about the issue.’ (lv + appl)(van den Berg & Busenitz 2012: 31)
56.3.2 Totoli
The Totoli voice and applicative system shows some similarities to that of Balantak. Himmelmann & Riesberg (2013) analyze Totoli as a Philippine-type language with three basic transitive constructions, av, pv, and lv, and two additional pivot-neutral applicatives. Like Balantak, Totoli does not show a cv basic voice category.
Applicative 1 in Totoli selects a beneficiary, instrument, or theme as a core argument and Applicative 2 selects a goal or recipient as a core argument (while lv selects a static location role as the syntactic pivot). In Himmelmann and Riesberg’s analysis, Applicative 1 and 2 may only occur with av and pv, never with lv. Example (123) shows the Applicative 1 construction with the verb taip ‘to peel’ and a beneficiary applied phrase.
(123) Totoli, Beneficiary-selecting construction
a.
I
Rinto
man-(t)aip
taipang.
hon
R.
av-peel
mango
‘Rinto peels mangoes.’ (av nonrealis, BC)
b.
I
Rinto
man-(t)aip-an
aku
taipang.
hon
R.
av-peel-appl1
1sg
mango
‘Rinto peels mangoes for me.’ (av nonrealis + appl1)
c.
Aku
ko-doong
pan-(t)aip-an
Rinto
taipang.
1sg
pot-want
sf-peel-appl1
R.
mango
‘For me, Rinto will peel a mango.’ (pv nonrealis + appl1)
d.
Aku
ni-pan-(t)aip(‑an)
Rinto
taipang.
1sg
rls-sf-peel(‑appl1)
R.
mango
‘For me, Rinto peeled a mango.’ (pv realis + appl1)(Himmelmann & Riesberg 2013: 400–401, 405)
Notably, there are some irregularities in the paradigm for the morphological marking of voice and applicatives in Totoli. The suffixal form ‑an, which is found with Applicative 1 constructions in av and pv nonrealis mode, is missing in pv realis mode in varieties of the language spoken in Totoli City, but included in the Northern Totoli dialect (Himmelmann & Riesberg 2013: 410–411). This is represented with parentheses around the suffix which I have added in example (123d). The gap in suffixal marking for Applicative 1 in pv realis most likely reflects pmp morphology for cv, which is marked by *-an in imperative/negative mood, and *Si-/*Sa- in indicative mood, with the prefix apparently lost in Totoli. Applicative 2 in Totoli is marked by ‑i except in pv realis, where it is marked by ‑an. This most likely reflects pmp morphology for lv, which is marked by *-i in imperative/negative mood and *-an in indicative mood. See § 64.2 and § 64.3 for more on development of pivot-neutral AMs from lv and cv morphology. The prefixal forms po(g)-/poN‑ (glossed as a stem-former above) are found on verbs in lv and certain Applicative 1 verbs, but not on Applicative 2 verbs.
In summary, Totoli has both a Philippine-type voice system with a productive lv alternation and pivot-neutral ACs that co-occur only with av and pv. Irregularities in the paradigm of morphological marking for voice and applicatives in Totoli suggest that reinterpretation of po(g)‑/poN‑, ‑an, and ‑i have played a role in the development of the pivot-neutral applicatives. In particular, morphological marking that was associated with cv and lv in earlier stages of development now co-occurs with marking for av; this would have been impossible in pmp. Thus, changes in the interpretation of some voice affixes in Totoli have given rise to the development of pivot-neutral applicatives.
56.3.3 Bobongko
In Bobongko, we see a different type of transitional system. The Bobongko voice system includes two major transitive voice alternations (Mead 2001). Verbs in av select the A argument as syntactically privileged, and are normally marked by moN‑ in realis mode, noN‑ in irrealis mode, and poN‑ in imperatives.18 Verbs in pv select the P argument as syntactically privileged, and are marked by o‑ in irrealis mode, ‑in‑ in realis mode, and no overt morphology in imperatives.
Bobongko also makes use of two suffixes that co-occur with av and pv morphology, which I will analyze as AMs: ‑i (and its variant ‑an), and ‑akon.19 The suffix ‑i/‑an marks the verb when a clausal argument is a semantic goal or location; in such constructions, the applied phrase appears to have the syntactic properties and coding typical of P in av and pv base constructions. In av, it normally appears as an unmarked NP immediately following the verb, as in (124a) below. In pv, the applied phrase is the apparent pivot or syntactically privileged argument as in (124b) below. Note that when the goal- or location-selecting construction appears in pv and realis aspect, the suffix ‑an always appears on the verb instead of ‑i.
(124) Bobongko, Locative-selecting applicative
a.
Jadi
no-sangalu
ka’a,
no-tugal
lampi’
so
rls-be.friends
this,
rls.av-plant
banana.
‘So they being friends, they planted bananas.’ (BC)(Mead 2001: 87)
b.
Ka’a-taa’
inaut
anu
t⟨in⟩ugal-an-ku
bele-nu
binte’.
this-that
garden
rel
⟨rls.pv⟩plant-loc.appl-1sg.gen
with-lnk
corn
‘This is the field that I planted with corn.’ (AC)(Mead 2001: 78)
The suffix ‑akon marks the verb when a clausal argument is a semantic beneficiary or recipient. However, unlike constructions marked with ‑i/‑an, in constructions marked with ‑akon, the beneficiary applied phrase is not realized with the syntactic properties and coding observed for P in av and pv base constructions. The beneficiary is instead marked with bele ‘with, for’, and does not appear to be a core argument. An example is shown in (125) below.
(125) Bobongko, Beneficiary-selecting applicative
a.
Gunsing
b⟨in⟩oa.
key
⟨rls.pv⟩carry
‘(Someone) has taken the key.’ (BC)(Mead 2001: 76)
b.
B⟨in⟩a-kon-nyo
bele-nu
sangalu-nyo,
anu
to-pomangan.
⟨rls.pv⟩bring-ben.appl-3sg.gen
for-lnk
companion-3sg.gen
rel
person-chew.betel
‘She brought (the lime) for his companions, who were betel chewers.’ (AC)(Mead 2001: 89)
In addition to av and pv, Bobongko makes use of constructions that Mead calls “special inverse” constructions. These appear to be of low textual frequency, though available data is limited. In one such construction, the instrument is syntactically privileged, and the verb is marked with the prefix poN‑ in irrealis mode and pinoN‑ in realis mode. This is shown in (126). In another, the location is syntactically privileged, and the verb is marked with the circumfix poN- -an in irrealis mode, and pinoN- -an in realis mode.20 This is shown in (127).
(126) Bobongko, Instrumental Voice
a.
Sapi’
taio’
kana’
o-kolot
cow
that
must
irr.pv-slaughter
‘The cow must be slaughtered.’ (pv)(Mead 2001: 76)
b.
Ka’a-mo
kapara
anu
ku-pong-kolot
sapi’
ka’a.
this-pfv
machete
rel
1sg.npiv.irr-irr.iv-slaughter
cow
this
‘Here is the machete with which I will slaughter the cow.’ (iv)(Mead 2001: 81)
(127) Bobongko, Locative Voice
a.
N-una’-ku-mo
kacamata.
rls.pv-stow-1sg.gen-pfv
eyeglasses.
‘I’ve put the eyeglasses away.’ (pv)
b.
Lamari
p⟨in⟩ong-una’-an-ku
cabinet
⟨rls.pv⟩lv-stow-lv.rls-1sg.gen
‘The cabinet is where I put them.’ (lv)(Mead 2001: 82)
As described thus far, it appears that Bobongko has a marginal Philippine-type voice system, with two major transitive voice alternations (av and pv), and two minor transitive voice alternations (iv and lv), in addition to pivot-neutral ACs.
On top of this, the instrument-selecting (iv) prefix poN‑ also co-occurs with the locative-selecting applicative suffix ‑i/‑an, shown in (128) below.
(128) Bobongko, Instrumental Voice + Locative-selecting applicative
a.
Oko
ku-bobal-i!
2sg
1sg.npiv.irr-pv.hit-loc.appl.irr
‘I will hit you!’ (pv + Locative-selecting AC)(Mead 2001: 78)
b.
Ka’a
kau’
anu
pom-bobal-i-nyo
ara
This
wood
rel
iv-hit-loc.appl.irr-3sg.gen
3pl
‘This is the wood he will hit them with.’ (iv + Locative-selecting AC)(Mead 2001: 81)
The semantic and syntactic properties of the type of clause shown in (128b) are consistent with co-occurrence of iv and the locative-selecting pivot-neutral applicative. Like the iv base construction, the instrument is always the pivot in this type of clause. Like the locative-selecting applicative in av and pv, the location or goal is a core argument, but not necessarily the pivot, as evident from its realization as a post-verbal unmarked free pronoun in (128b). The semantic role selected as pivot depends on the basic voice alternation.
In contrast, I do not analyze the construction marked with poN- -an/pinoN- -an, shown in (127) above, as co-occurrence of instrumental voice (iv) and the locative-selecting applicative marked with ‑i/‑an. This construction does not share semantic or syntactic properties that are characteristic of the iv base construction; an instrument is not seen to be part of the semantic meaning of the clause and is not the role selected as pivot. Furthermore, the locative-selecting applicative is marked with ‑an only in realis mode for pv, and ‑i everywhere else. But the construction in (127) always shows a suffixal ‑an component, both in realis and irrealis mode, according to Mead (2001: 83).
Finally, in a single example, we observe a construction marked with pinoN- -akon-an in realis mode, shown in (129).
(129) Bobongko, Beneficiary as pivot
Aliali-um
anu
p⟨in⟩om-be’-akon-an-ku
kapara-um
younger.sibling-2sg.gen
rel
⟨rls⟩lv-give-ben.appl-lv-1sg.gen
machete-2sg.gen
‘It was your younger sibling to whom I gave your machete.’(Mead 2001: 83)
This construction is also ditransitive, taking two nonactor core arguments coded as unmarked NPs, a recipient, and a theme. The beneficiary or recipient, aliali-um ‘your younger sibling’, is the pivot (R), while the theme, kapara-um ‘your machete’, is a non-pivot core argument (T) appearing immediately following the verb. This type of clause could be analyzed as a minor nonactor voice, that is, a bv construction, marked with pinoN- -akonan. Alternately, it could be analyzed as co-occurrence of lv, marked with pinoN- -an, and the beneficiary-selecting applicative, marked with ‑akon.
Evidence for analysis as bv includes the fact that a location is not the pivot, and therefore this construction does not share a key property of lv base constructions. Evidence for analysis as lv plus pivot-neutral applicative ‑akon includes the fact that this clause is ditransitive, while other instances of ‑akon in co-occurrence with av and pv are not. In this second possible analysis, we might consider the function of ‑akon as adding a semantic beneficiary or recipient to the constructional meaning, and the function of pinoN- -an to be selecting the most location- or goal-like argument as the pivot—that being the semantic recipient rather than the semantic theme. Given the limited data, a definitive assessment is not possible, though I lean towards analyzing this construction as lv plus a co-occurring pivot-neutral beneficiary-selecting applicative. Interestingly, the ‑akon suffix appears closer to the verb stem than does the suffixal portion of the circumfix pinoN- -an. This is consistent with analysis in which a beneficiary or recipient is added first, perhaps at a semantic level of verbal structure, and is subsequently selected to be realized as the syntactic pivot.
The paradigm of morphological marking for symmetrical voice alternations and pivot-neutral applicatives in Bobongko, as analyzed here, is shown in Table 31. This system has four Philippine-type voice alternations and two pivot-neutral ACs, plus an aspectual distinction marked with ‑in‑ (or its variant n‑) for realis mode. The two pivot-neutral ACs co-occur with both major voice categories (av and pv), but each is found with only one of the two minor voice alternations, that is, either iv or lv, but not both.
Bobongko thus represents one type of transitional system that shows Philippine-type voice as well as pivot-neutral applicatives. However, iv and lv appear at much lower textual frequency and appear to be in the process of being lost. In a pure Philippine-type voice system, each peripheral nonactor semantic role, e.g. beneficiary, instrument, location, etc., can be selected only as the pivot by the use of voice morphology on the verb, that is, by means of what I have called pivot-selecting ACs. In Bobongko, we see the emergence of pivot-neutral ACs in which such peripheral roles may alternately be selected to map to the pivot, a non-pivot core argument, or an oblique, depending on the combination of co-occurring constructions. There are some irregularities and gaps in the Bobongko paradigm; such as the marking of pv with o‑ only in irrealis mode and not realis mode. There are also differences between the locative-selecting applicative marked with ‑i/‑an vs. the beneficiary-selecting applicative marked with ‑akon; each co-occurs with a different set of nonactor voices, and shows different patterns in the syntactic properties of the applied phrase across these voices. Such irregularities might be expected when a language exhibits a transitional state, as I am arguing that we see here, with the loss of Philippine-type voice and the development of pivot-neutral applicatives both in progress.
Table 31
Partial paradigm for voice and applicatives in Bobongko
|
Voice |
Mode |
Base |
Loc. Appl. |
Ben. Appl. |
|---|---|---|---|---|
|
av |
irrealis |
moN- |
moN‑ +‑i |
moN‑ + ‑akon |
|
av |
realis |
noN- |
noN‑ + ‑i |
(no data) |
|
pv |
irrealis |
o- |
o‑ + ‑i |
o‑ + ‑akon |
|
pv |
realis |
‑in- |
‑in‑ + ‑an |
‑in‑ + ‑akon |
|
iv |
irrealis |
poN- |
poN‑ + ‑i |
— |
|
iv |
realis |
pinoN- |
pinoN‑ + ‑i |
— |
|
lv |
irrealis |
poN- -an |
— |
(no data) |
|
lv |
realis |
pinoN- -an |
— |
pinoN- -an + ‑akon |
Note: The indication ‘no data’ is used when the form is predicted to be possible but is not found in the given mode in available data. The indication ‘—’ is used when there is no evidence that the form is possible in either mode.
The exact type of system as found in Bobongko is rare or even unique in West Nusantara. However, there are a number of languages in the sample that show “special” ACs akin to certain Bobongko lv and iv forms, where an applicative suffix appears together with some other fossilized affixal marking on the verb under certain conditions. One example is Pendau, which is discussed in the following section.
56.3.4 Pendau
Pendau is a Tomini language spoken in the northern half of Central Sulawesi province. Like Bobongko, Pendau shows two major voice alternations (av, pv) and two applicative suffixes, ‑a’ and ‑i. A partial paradigm for symmetrical voice alternations and ACs in Pendau is given in Table 32.
Table 32
Partial paradigm for voice and applicatives in Pendau
|
Class |
Voice |
Mode |
Base |
Goal/loc. appl. |
Ben./Inst. appl. |
|---|---|---|---|---|---|
|
Prim. Trans. |
av |
irrealis |
M‑ + pong- |
M‑ + pong‑ + ‑i |
M‑ + pong‑ + -a’ |
|
Factive |
av |
irrealis |
M‑ + po- |
— |
M‑ + po‑ + ‑a’ |
|
Prim. Trans. |
av |
realis |
N‑ + pong- |
M‑ + pong‑ + ‑i |
M‑ + pong‑ + -a’ |
|
Factive |
av |
realis |
N‑ + po- |
— |
N‑ + po‑ + ‑a’ |
|
Prim. Trans. |
pv |
irrealis |
ro- |
ro‑ + ‑i |
ro‑ + ‑a’ |
|
ro‑ + pong‑ + ‑i |
ro‑ + pong‑ + ‑a’ |
||||
|
Factive |
pv |
irrealis |
ro‑ + po- |
— |
ro‑ + po‑ + ‑a’ |
|
Prim. Trans |
pv |
realis |
ni- |
ni‑ + ‑i |
ni‑ + ‑a’ |
|
ni‑ + pong‑ + ‑i |
ni‑ + pong‑ + ‑a’ |
||||
|
Factive |
pv |
realis |
ni‑ + po- |
— |
ni‑ + po‑ + ‑a’ |
Note: The indication ‘no data’ is used when the form is predicted to be possible but is not found in the given mode in available data. The indication ‘—’ is used when there is no evidence that the form is possible in either mode.
In Pendau, a so-called ‘stem former’, pV(C)‑ appears on the verb in certain constructions. The exact form of the stem former is lexically determined by verb class; for example, most transitive verbs take pong‑, while factive verbs (e.g. ‘make’, ‘build’) take po‑, dynamic verbs (e.g. ‘search’) take pe‑, and postural verbs (e.g. ‘sit’) take popo‑. With certain classes of verbs, such as factives, the stem former is always obligatory in finite verb forms. With others, such as the primary transitive and dynamic verb classes, the stem former is generally absent in pv.
The appearance of the stem-forming prefix shows irregularities in certain ACs with primary transitive verbs (bolded cells in Table 32). In ACs marked with ‑a’ that take beneficiary applied phrases, the stem-forming prefix may be absent in pv forms, as is the usual pattern for this verb class. But in ACs marked with ‑a’ that take instrument applied phrases, the stem-forming prefix always appears. So, the pv verb form ni-pong-gabu-a’ ‘cooked with’ takes an instrument applied phrase while ni-gabu-a’ ‘cooked for’ takes a beneficiary applied phrase. Similarly, in ACs marked with ‑i that take goal applied phrases, the stem-former prefix may be absent in pv forms. But in ACs marked with ‑i that take static location applied phrases, the stem-forming prefix is obligatory.
Furthermore, for both primary transitive and factive verbs, instrument and static location applied phrases are always realized as the pivot, and only found when the verb is morphologically marked like other pv constructions, i.e. with ro‑ (irrealis) or ni‑ (realis). Beneficiary and goal-selecting ACs, on the other hand, are compatible with both the pv and the av constructions.
I argue that the use of pV(C)‑ with primary transitive verbs in instrument- and static location-selecting ACs in Pendau represents residual marking for iv and lv constructions.21 In Bobongko, there are formal differences in morphological marking for base forms in lv, base forms in iv, and pv + applicative suffixes. In Pendau, such distinctions are minimal. Because the pV(C)‑ morpheme has no clear function in many verbal forms, and its presence in pv is only unusual for certain verb classes, its presence in the special applicatives looks like an irregular use of the stem former rather than part of distinct affixes marking separate iv and lv constructions. In most verb classes in Pendau, the form of the verb in pivot-neutral goal-selecting ACs in pv is identical to the verb in the pivot-selecting locative constructions, and the form of the verb in pivot-neutral benefactive ACs in pv is identical to the verb in the pivot-selecting instrumental ACs. Still, there is admittedly a degree of irregularity and ambiguity in the paradigms, that allow for different analyses of both Bobongko and Pendau.
In summary, Pendau shows remnants of iv and lv constructions in irregular clausal properties and morphological marking for instrument-selecting and locative-selecting ACs, as compared to beneficiary-selecting and goal-selecting ACs. Like Balantak, Totoli, and Bobongko, in Pendau, reduction in Philippine-type voice categories coincides with development of pivot-neutral applicatives. Together, these four cases indicate that the transition between Philippine-type voice and pivot-neutral applicatives in West Nusantara may take place via (i) paradigmatic reduction of cv and/or lv voice categories, (ii) reinterpretation of cv and/or lv affixal marking as applicative markers or stem formers that may appear on verbs in formerly contrastive voice categories (or functional replacement of such affixal marking with new AMs), and (iii) emergence of productive use of the new applicative markers in co-occurrence with pv, and crucially, av transitive constructions, such that a selected peripheral role may map to the pivot argument, or to a non-pivot argument of the clause, depending on voice.
57 Summary of Major Findings
This chapter has explored the distribution of languages of West Nusantara with and without applicatives with close attention to the influence of geography and typological profiles associated with particular geographic locations. Looking at geographic and typological factors together helps to explain which languages in the sample show continuity with earlier diachronic stages by retention of the Philippine-type voice system, which lost this continuity and lack applicatives altogether, and which show change by developing pivot-neutral applicatives.
Retention of the Philippine-type voice system, in which the peripheral nonactor voices (lv, cv) constitute pivot-selecting ACs, is found primarily in and adjacent to the Philippine archipelago, and mainly in the Sabahan languages and the Sama-Bajaw languages (Greater Barito), with some occurrence in other languages of northern Sulawesi. Thus, languages of West Nusantara, like other western Austronesian languages, generally show decay in the Philippine-type voice system as one moves outside of Taiwan and the Philippines to the south (see McDonnell & Chen 2022).
Pivot-neutral applicatives that co-occur across basic voice alternations, on the other hand, are found to be a broadly distributed areal feature for West Nusantara associated with loss or reduction of the four-way Philippine-type voice system. Such applicatives are even found in languages (e.g. Mentawai, Enggano) which had little to no contact with Malay and other influential languages of the region. The presence of pivot-neutral applicatives also cuts across languages with a wide range of structural features: symmetrical and asymmetrical voice systems, and different types of morphological and syntactic alignment, systems of case-marking, and word order patterns. Given this, I conclude that there is no cohesive typological profile for languages of West Nusantara showing applicatives that co-occur across basic voice categories, by which they may be distinguished from West Nusantara languages without such applicatives.
Considering these findings, the lack of applicatives, rather than the presence of them, is a pattern that requires explanation in languages of West Nusantara. Based on the survey results and additional examination of descriptive resources for languages of West Nusantara outside of the sample, two profiles were developed to summarize key typological features of the languages of the sample without applicatives. One of these is centered geographically on languages of mainland Southeast Asia and northern Peninsular Malaysia, and the other on Bornean languages south of Sabah. For both sets, loss of suffixation appears to be a major factor in the absence of applicatives. We also observe general reduction of morphological complexity, a greater reliance on word order to signal grammatical relations, and a shift towards analytic structures as alternatives for certain voice and applicative alternations observed elsewhere in West Nusantara. In the mainland Southeast Asia set, these changes are more complete across the grammars of individual languages, especially those with a greater time depth of language contact with non-Austronesian languages. In the Bornean set, these changes might affect different verbal constructions, e.g. av and pv, to different extents and in different manners. Because of this, the Bornean set is less cohesive compared to the mainland Southeast Asia set. Due to this fact, plus the existence of gaps in our understanding of the history of various Bornean language groups and lack of basic linguistic description for many others, the profile developed for the Bornean languages lacking applicatives is more preliminary in nature and less certain.
On the whole, the findings presented in this chapter cast further doubt on the appropriateness of an Indonesian-type subcategory of western Austronesian symmetrical voice languages, defined by two- or three-way symmetrical voice alternations, applicatives, and use of special clitic pronominal forms for indexing non-pivot actors (Himmelmann 2005: 175). While many of the better known languages of the region meet these criteria, e.g. Standard Indonesian, Balinese, and Javanese, when we consider a larger, more representative sample of languages of West Nusantara, other conflicting combinations of features are quite broadly attested. Many Bornean languages cross-cutting a diverse number of lower-level genetic groupings have no applicatives, but otherwise show features consistent with the definition of Indonesian-type languages. Conversely, outside of Borneo and mainland Southeast Asia, languages with voice and alignment systems with diverse characteristics are all found to have pivot-neutral applicatives in the great majority of cases.
In conclusion, pivot-neutral applicatives that co-occur across basic voice categories should be considered an areal feature of languages of West Nusantara, though such applicatives may be lost or blocked from developing due to special factors. These include phonological and/or morphological changes affecting word structure, which may be triggered or spread by language contact. Furthermore, evidence from broad patterns of distribution, as well as nuanced interpretation of descriptive data in individual languages, show that this type of applicative arises in the face of devolution of the Philippine-type voice system.
In the next chapter, I look closer at distributional patterns in the forms and functions of AMs and ACs in the survey, and argue that many AMs marking pivot-neutral applicatives are in fact derived from AMs that marked Philippine-type voice alternations (lv, cv) in pmp and pan. The distributional patterns for pivot-neutral applicatives thus represent both continuity with earlier diachronic stages and specific types of functional change.
Acehnese is the only Chamic language spoken primarily outside of mainland Southeast. Unlike other Chamic languages (see §54), Acehnese does have one construction that may be considered an applicative: when the verb is marked with the prefix peu‑, the clause may mean ‘administer root onto/at an undergoer’ (Durie 1985: 79). This construction meets the definition of applicative used in this study because it selects a goal as a core argument. However, most other meanings associated with peu‑ are causative and not applicative, and the applicative usage is of limited productivity.
In main clauses and most dependent clauses, Nias does not show an alternation between A-oriented and P-oriented transitive constructions. A relic of symmetrical voice is found only in certain P-oriented relative clauses marked with ni‑, in which the actor is frequently overtly expressed (see Brown 2001: 421).
Sigulai probably also represents an asymmetrical voice system, while Mentawai appears to be a two-way symmetrical system (see McDonnell & Truong 2024b: 417–420).
The discussion of Enggano here is based on Crowley n.d., which relies on data from Kähler (1940). Newer documentary and descriptive work on contemporary Enggano is becoming available, including a forthcoming edited volume (Arka, Hemmings & Dalrymple in prep). Nonetheless, the older source material for Enggano is informative as it reflects the language as spoken before increased language contact with Malay/Indonesian and other languages of Sumatra.
The proclitic construction is reported to be rare with second person non-pivot actors and relatively uncommon with first person actors in Tengger, probably due to avoidance of these forms for politeness (Conners 2008: 146–147).
Patterns for the third person proclitic differ slightly, as it is used only in transitive clauses, and thus only for A (Shiohara & Arka 2024).
The particle marking the A argument may also be realized as gemination on the initial consonant of the NP expressing A.
Tatana is considered an unclassified Sabahan language by Eberhard, Simons & Fennig (2021), but alternately, may subgroup with Murutic (Hammarström et al. 2022).
The construction that I call locative voice (lv) here is called “setting voice” by Dillon (1994). It should not be confused with the separate locative adversative construction marked with ‑on in Tatana. The locative adversative is not a basic voice category in Tatana (see Dillon 1994: 61–64).
Intransitive av shows distinct marking from transitive av in Tatana. For the sake of simplicity, the former is not included in Table 24.
Suffixation is not found in the four Land Dayak languages of the sample. Matéq shows a discourse marker =éh, which behaves as a clitic. There is also no suffixation in Bidayuh Serian [sdo] (Omar 1983: 447–451), Biatah Bidayuh [bth] (Omar 1983: 446–469), nor in Sungkukng and other Bidayuhic languages in the Bengkayang, Landak and Sanggau regencies of West Kalimantan, Indonesia (Adelaar 2006: 81, citing personal fieldnotes). Rensch et al. (2012) find only enclitics, prefixes, and a few infixes, some of which are fossilized, in their study of over 25 Bidayuh varieties.
For Punan Tubuh, Soriente (2013: 184) makes brief mention of a possible benefactive function for the verbal prefix pe‑. However, this is not illustrated with lexical or clausal examples by which it can be verified that clauses marked with this morpheme meet the definition of an AC used in this study. I have classified Punan Tubah as lacking applicatives for this reason.
This construction may not necessarily be an innovation, but its frequent use in place of morphologically marked pv, especially in indicative clauses, does appear to mark a diachronic shift. The form of the verb meaning ‘do/make’ in North Sarawak languages is often an, or similar, which is possibly a reflex of a preverb *an, held to be used in cv in non-indicative moods in the pan voice system by Ross (2009). For more details, see the discussion in §63.6.
Alongside productive av and pv constructions, Duri shows another A-oriented ‘antipassive’ construction with reduced semantic transitivity and very limited textual frequency (Valkama 1993, see also discussion in §73).
In most South Sulawesi languages, A of A-oriented constructions is indexed as an enclitic form (or suffix) on the verb and P is not indexed. If A-oriented clause are syntactically intransitive, this is consistent with ergative alignment patterns, though see below for some reasons that the transitivity value of these clauses is not straightforward. Furthermore, in some languages, the core argument of an intransitive verb (S) may show different patterns of indexing in certain irrealis or non-indicative clauses, complicating patterns of morphological alignment for core arguments (see D. Laskowske 2016: 43–47; K. Laskowske 1994, B. Friberg 1991: 120–121).
Languages coded as asymmetrical for voice in Bungku-Tolaki and Muna-Buton, do generally show more than one (apparently) transitive construction, e.g., “ae-class marking” vs. “a-class marking” in Muna (van den Berg 1995), and zero-marked verbs vs. moN‑ marked verbs in Moronene (Andersen & Anderson 2005). These languages are not coded as two-way symmetrical voice systems because there is little evidence that A in one such construction is syntactically privileged, while P is syntactically privileged in the other. To the contrary, these languages generally show syntactic behavior indicating that S and A constitute a privileged syntactic relation to the exclusion of P (accusative alignment). Alternations between transitive constructions in these languages are often distinguished by definiteness constraints on P and semantic transitivity, in which sense they are similar to clausal alternations described above for South Sulawesi languages (§56.1).
In addition to a transitive or ‘agented’ pv construction in which A may be overtly expressed, Balantak also has a distinct agentless P-oriented construction, in which the actor argument is suppressed (see van den Berg & Busenitz 2012: 29).
What Mead (2001) describes as an aspectual distinction between realized and unrealized aspect in Bobongko appears to correspond to realis and irrealis mode (or nonrealis mode) in Balantak, Totoli, Pendau, and other Philippine-type languages. In all four languages discussed in this section, this distinction encompasses differences in both tense/aspect and evidentiality/factuality. In Bobongko examples, the realis or realized aspect forms are used to describe past completed actions and states, while the irrealis or unrealized aspect forms are used to describe ongoing and future actions and states, and in certain requestive and prohibitive constructions. For the sake of consistency, I have used realis [rls] and irrealis [irr] to distinguish these forms in discussion and glossed examples for Bobongko.
Mead 2001: 77 does not refer to ‑i (or its variant ‑an) as AMs, primarily because its compatibility with verbal roots is almost entirely lexically-conditioned. Nonetheless, constructions marked with ‑i/‑an in Bobongko meet the definition of applicative used in this study.
There is no available data on imperatives in these two constructions in Bobongko.
Instrumental and location nominalizations in Pendau are also formed with the pV(C)‑ prefix, and this occurs across several verb classes, including a number for which ACs of this type are not attested (Quick 2007: 99–100).