Non-productive antipassive in non-ergative languages
Merel Hendriks
Student Number: 11294523
Programme: General Linguistics, University of Amsterdam Supervisor: dr. Eva van Lier
2
Table of Contents
Abbreviations ... 3 1. Introduction ... 4 2. The Antipassive ... 5 2.1 Origins... 5 2.2 Definition ... 52.3 Transitivity and Valency ... 7
2.4 Productivity ... 8 3. Method ... 11 4. Results ... 12 4.1 Ainu... 13 4.2 Haida ... 16 4.3 Lunda ... 18 4.4 Koyraboro Senni ... 19 4.5 Lango (Uganda)... 21 4.6 Mocoví ... 23 4.7 Tundra Nenets ... 25 4.8 Tundra Yukaghir ... 26 4.9 Neverver ... 27 4.10 Irarutu ... 28 4.11 Collective Results ... 30 5. Discussion ... 31 5.1 Main Findings ... 31
5.2 Limitations and future ... 33
5 Conclusion ... 34
References ... 35
3 Abbreviations
COMP Comparative FIN/FP Final particle
HAB Habitual
LOGO Logophoric Pn Plural inclusive
PP Postposition
PROL Prolative
4 1. Introduction
The focus of this thesis is on non-productive antipassives in languages with a non-ergative alignment system. The antipassive construction is a derived construction, where the patient-like argument is suppressed or demoted. It is commonly associated with language with an ergative alignment system, however it has been shown to occur in accusative languages.
The productivity and the limitations placed on verbs that allow the antipassive construction, especially in accusative languages, have been little studied. Polinsky (2013) is one of few large-scale studies on the productivity of the antipassive. Say (Forthcoming) is an exploratory study on the restrictions of the antipassive construction. Say reproduces a hierarchy based on a verbs’ ability to allow the antipassive construction and proposes five semantic properties that increase the likelihood of a verb allowing the antipassive.
In this thesis, the antipassive verbs of ten ergative languages with
non-productive antipassives are compared to the hierarchy and the properties as mentioned in Say (Forthcoming).
In the chapter The Antipassive, the origins of the term antipassive, the definition of the antipassive, transitivity and valency, and the productivity of the antipassive, the
hierarchy and the properties will be discussed. In the chapter Method, the methodology will be presented. Here, the criteria that the languages had to satisfy and how the languages were compared to the hierarchy and the properties, is discussed.
In the chapter Results, the background information and the antipassives of the 10 languages will be given. This is followed by the comparison of the verbs that allow the antipassive in the languages.
Following the Results, is the Discussion. Here the data presented in the Results chapter will be examined further and interpreted. Some of the limitations will be discussed. Finally, a conclusion will be given, which provides a brief overview of the findings of this thesis.
5 2. The Antipassive
2.1 Origins
The term ‘antipassive’ was originally created by Michael Silverstein (1972) in his description of Chinook Jargon. The antipassive construction was originally used by Silverstein, as a construction in ergative languages. Silverstein described a construction where a transitive structure becomes intransitive with an implied object through the use of the morpheme -ki-. The antipassive construction was seen by Silverstein as an “inverse equivalence” of the passive in accusative languages (Silverstein, 1972: 395).
Silverstein does not provide clearly glossed examples and not every segment has been given a corresponding gloss, but to give some extra clarity on the structure example 1a-b is given. In example 1b, the use of the -ki- morpheme is shown. Regarding example 1a, Silverstein describes k-ɑ-ʎɑ́-k’ɑuk’ɑu as the ergator and ʎ-k’ɑsks as the non-ergator. Each has an “inflectional pronominal in apposition in the verb – a transitive verb” (Silverstein, 1972: 395). In example 1b, the directional -u- from 1a is replaced by -ki-.
(1a.) ɑ-ʎk-ʎ-ú-kšt-x ʎ-k’ɑsks k-ɑ-ʎɑ́-k’ɑuk’ɑu
he-looks-at-it-habitually child one-who-‘shot’-people (transl. murderer) (1b.) ɑ-ʎ-ki-kšt ʎ-?xɑt ʎ-kwɑʎílxmwk
he-looked one person
Silverstein (1972: 395)
2.2 Definition
The antipassive construction is commonly referred to as detransitivizing process, where the patient or patient-like argument is suppressed or demoted1. The patient-like argument can be realized as an oblique phrase or left implicit. As observed by Silverstein, the antipassive is in a sense the opposite of the passive construction, where the agent-like argument becomes oblique or is left implicit.
The transitive/antipassive alternation is shown with examples from two languages below. In example 2, the antipassive in Chukchi is shown and in example 3 an example from Dyirbal is shown. In 2a , the ergative A and absolutive O agree with the transitive verb ‘carry’. In 2b, the antipassive prefix ine- is used. The A-argument is now expressed with an
6 absolutive case. The verb no longer agrees with the P-argument, which is now expressed with an oblique case.
(2a.) ʔaaček-a kimitʔ-ən ne-nlʔetet-ən
youth-ERG load-ABS 3PL.SUBJ-carry-AOR.3SG.OBJ ‘The young men carried away the/a load.’
(2b.) ʔaaček-ət ine-nlʔetet-gʔe-t kimitʔ-e youth-ABS ANTIP-carry-AOR.3SG.SUBJ-PL load-INSTR ‘The young men carried away the/a load.’
Polinsky (2013) Example 3a shows the use of the transitive verb ‘see’. In 3b, the antipassive is
indicated by the use of a verbal marker and a dative case marker on the patient. This
suggests that the patient is not a core argument of the verb. The ergative case marker of the agent has also been changed into an absolutive form.
(3a.) yabu ŋuma-ŋgu bura-n mother.ABS father-ERG see-NFUT ‘Father saw mother.’
(3b.) ŋuma bural-ŋa-nyu yabu-gy
father.ABS see-ANTIP-NFUT mother.DAT ‘Father saw mother.’
Heaton (2020: 25) There exists debate on what exactly qualifies as an antipassive. Some consider the antipassive to be linked to ergative languages (Silverstein, 1972, 1976; Cooreman, 1994; Palmer, 1994; Aldridge, 2007), others argue that the antipassive also presents itself in accusative languages (Polinsky, 2013, 2017; Janic, 2016; Heaton, 2017, 2020; , Dom &
Segerer, 2015). Polinsky (2017: 19) argues that visibility of the antipassive is the cause of this issue. The antipassive construction is more noticeable in, but not limited to, ergative
languages, as the presence of the antipassive with a change in subject encoding (ergative to absolutive, see example 1b). The focus of the current paper is on the antipassive in non-ergative languages, and as such it is assumed that the antipassive can occur in
nominative/accusative languages.
The antipassive is generally taken to be a detransitivizing construction (Vigus, 2018; Bostoen et al., 2015). Some refer to the antipassive as a valency-reducing construction as
7 well (Heaton, 2017: 7). This can be caused by differences in the definition of valency and transitivity. Due to this, transitivity and valency will be discussed below in 2.3.
2.3 Transitivity and Valency
Whether the antipassive is a intransitivizing construction, or valency-decreasing
construction, depends on the different definitions of valency, and the sometimes unclear distinction between transitivity and valency.
Transitivity. Heaton (2017: 3) defines transitivity as a property of an entire clause, consisting of a verb and its arguments. Transitivity can be reduced and this reduction is often indicated by morphosyntactic means, for example a change in case. Heaton adopts a scalar view of transitivity, where changes in transitive morphosyntax commonly reflect a decrease in transitive features. Through this definition, the antipassive can be described as
detransitivizing.
Valency. Heaton (2017: 7) refers to the antipassive as both a valency-reducing and detransitivizing construction. Here, valency means the number and kind of arguments that a particular predicate can take. Others state that valency refers to a particular valency pattern (Kulikov, 2010) , or to a verbal category which marks a change in valency. Heaton discusses transitivity as a scalar concept which is grammatically categorical in some languages. Valency is discussed as a “property of lexical units in relation to constructions”, meaning that the verb’s valency is reliant on the construction it is in (Heaton, 2017: 9). Due to these
definitions, the antipassive can be considered both detransitivizing, and valency-reducing.
According to Kulikov (2010: 370), voice deals with the mapping of semantic arguments onto syntactic functions. A modification of voice only suggests changes in the mapping, but not the semantics of the sentence. No semantic role is removed from or added to the base structure. Kulikov refers to the antipassive as a type of voice that does preserves the semantic roles present in the base structure. This does not exclude roles that are
unexpressed, as their presence is implied by the meaning of the sentence (Kulikov, 2010: 374). This definition of voice would make the antipassive a detransitivizing construction, but not a valency-reducing construction.
8 2.4 Productivity
Not much research has been done on the productivity of the antipassive. Janic and Witzlack-Makarevic (Forthcoming: 36) state that Polinsky’s (2013) study is the only large-scale study on the productivity of the antipassive construction. In this study, Polinsky sampled 186 languages. Of these languages, 146 did not have an antipassive. Of the 40 languages that did have an antipassive, 24 had a productive antipassive, 14 a partially productive antipassive and 2 had a non-productive antipassive. However, Polinsky does not mention the criteria used to classify the productiveness of the antipassive in the sampled languages. Both Janic and Witzlack-Makarevic (Forthcoming) and Say (Forthcoming), state that lexical restrictions on the antipassive have been typically discussed for individual languages.
Say’s (Forthcoming) study is an exploratory paper aimed examining the typological generalizations that are underlying to the lexical restrictions placed on verbs that
exceptionally (dis)allow the antipassive construction. Say (Forthcoming) discusses a class of verbs that he refers to as “natural antipassives”. These are verbs that crosslinguistically are most easily compatible with the antipassive construction. Say (Forthcoming: 5) identifies five semantic properties that are typical of natural antipassives, based on previous typological studies:
1. agentivity of the A-argument; 2. specification of the A’s manner;
3. inherent atelicity (with the possibility of compositional telicity in the verb’s transitive use);
4. narrow class of potential P-arguments; 5. affectedness of the A-argument
The first property refers to the Agent having animacy, volitionality and control. This is not a restrictive property, in the sense that agentivity is a typical feature for most transitive verbs. Transitive verbs that select inanimate A-argument or A-arguments lacking control, such as ‘drop’ and ‘surprise’, are not natural antipassives.
The second property refers to idea that natural antipassives are typically manner verbs, rather than result verbs. Manner verbs lexically specify the manner in which the verb’s is carried out, such as ‘wash’ and ‘wipe’, whereas result verbs specify the result of the verb’s action, such as ‘break’ and ‘build’. A complication here is that some verb with a result
9 component can easily allow the antipassive construction, like the verb ‘eat’. This is possibly due to the antipassive meaning of the verbs highlighting the verb’s manner component. This means that the presence of a manner component is part of the second property, but that the possibility of a result component cannot be ruled out.
Property three is based on the theory that atelic verbs are more likely to allow the antipassive construction than telic verbs. Often the antipassive construction is atelic, whereas the transitive construction is telic. The verbs belonging to the natural antipassive are those where the telicity can be cancelled in the antipassive construction and can then be interpreted as an activity, as in eat apples. The telicity of the verb in the transitive
construction is compositional.
The fourth property means that the antipassive is more likely to apply to verbs that can select patient arguments from a relatively narrow semantic class. This includes verbs like ‘eat’ and ‘shave’. Transitive verbs that can apply to many different types of patients, such as ‘see’, do not enter the antipassive construction as easily.
The final property is the affectedness of the A. The A-argument is in some way affected by the action or event expressed by the verb, such as in ingestive verbs (‘eat’, ‘drink’), where the A is the causer and the endpoint, and in mental verbs where the A can be also thought of as an endpoint (‘hear’, ‘read’).
These five properties are not all necessary for a verb to allow the antipassive
construction. They are however considered contributing factors to the likelihood that a verb will behave as a natural antipassive.
The 80 verbs listed in the Valency Patterns Leipzig Online Database (ValPaL; see Hartmann, Haspelmath, & Taylor, 2013) and Malchukov’s (2005) hierarchy for object-demoting/deleting alternations were the basis for the hierarchy (Say, Forthcoming: 4). The hierarchy is a variant of Malchukov’s hierarchy. The hierarchy is based on the abilities of the verbs to enter antipassive derivations. Say notes that this ranking does not discriminate between object-demoting and object-deleting antipassives. The ranking is shown in 4:
(4.) ‘eat’, ‘shave’, ‘give’, ‘think’, ‘steal’ > ‘wash’, ‘cut’, ‘take’, ‘cover’, ‘wipe’, ‘see’, ‘search for’, ‘hit’, ‘throw’, ‘hear’ > ‘cook’, ‘know’, ‘ask for’, ‘tell’ > ‘beat’, ‘tear’ > ‘pour’ > ‘fill’, ‘climb’, ‘hug’, ‘look at’, ‘help’, ‘name’ > ‘break’, ‘kill’, ‘touch’, ‘load’, ‘teach’, ‘smell’ > ‘fear’, ‘dress’ > ‘show’, ‘send’, ‘carry’, ‘tie’, ‘put’ > ‘sing’, ‘grind’, ‘dig’ > ‘follow’, ‘say’, ‘build’, ‘peel’ > ‘jump’, ‘like’, ‘shout at’, ‘leave’, ‘live’, ‘play’, ‘meet’, ‘talk’, ‘hide’ > ‘blink’,
10 ‘laugh’, ‘roll’, ‘burn’, ‘frighten’, ‘run’, ‘be dry’, ‘push’, ‘bring’ > ‘cough’, ‘sit’, ‘go’, ‘scream’, ‘feel pain’, ‘sink’, ‘be a hunter’, ‘boil’, ‘sit down’, ‘die’, ‘be sad’, ‘feel cold’, ‘be hungry’, ‘rain’.
The relatively limited amount of research on the antipassive in non-ergative languages and the lexical restrictions placed on the antipassive in these languages, motivated the topic of this study.
11 3. Method
In this study, 10 languages with a non-productive antipassive construction are sampled. The languages were found through existing literature on antipassives, specifically typological studies (Bostoen et al., 2015; Heaton, 2017, 2020; Polinsky, 2017; Vigus, 2018) and an edited volume on antipassives (Witzlack-Makarevich & Janic, Forthcoming). The sampled languages had to satisfy a couple of criteria. First, they had to be non-ergative languages. This means that languages with split ergative alignment could not be included, but languages with active, accusative or mixed (more than one type of non-ergative) alignment could.
Second, the antipassive had to be non-productive, or at least not very productive, meaning that the antipassive cannot be applied to a large number of or to all transitive verbs. Productivity was based on the limitations mentioned in previous literature, or in the grammars. The sampled languages were discussed in the literature, mentioned above, as non-productive or were mentioned to have limitations on the application of the antipassive construction.
The antipassive also had to be marked on the verb in some way. This marking does not need to be dedicated exclusively to the antipassive. It is for example not uncommon for the marker to also apply to the anticausative or the reflexive/reciprocal. Heaton’s (2020: 202) sample suggests that most languages (86 out of 126) do not have a dedicated antipassive marker.
After the languages were selected and their relevant information was collected from their grammars, the individual verbs that allowed the antipassives were written down (see table 3). A different amount of verbs listed per language was anticipated, as the details of description of the grammars differ. Some might list only a few of the verbs that allow the antipassive, others might provide a more comprehensive list.
The verbs and the types of verbs that allow the antipassive were compared to the five semantic properties of “natural antipassives” and the hierarchy discussed in Say
(Forthcoming). These semantic properties and the hierarchy are listed earlier in 2.4. When the type of verbs that allow antipassives or their features were discussed, but not the specific verbs, these descriptions were compared to the semantic properties mentioned above.
12 4. Results
In this section, a brief overview will be given of all the sampled languages. Subsequently, the languages and their antipassive will be discussed individually. In the language-specific
segments, first information on the genealogy, geography, the speakers and some typological features will be provided, after which their antipassive construction will be discussed.
After the languages are discussed individually, the collective results will be covered. Here the verbs that appear in more than one grammar will be discussed, and compared to the properties and hierarchy.
In table 1, shows an overview of all the ten languages. The language families, the regions, the alignment systems based on agreement and case, and the language-specific sources are given. A mixed alignment denotes a different alignment pattern for case and agreement. Most languages belong to different language families, except for Neverver and Irarutu. Both these languages belong to the Austronesian language family, though from different subfamilies.
Table 1. Overview of Selected Languages
Language Language Family Region Alignment Sources
Ainu Isolate Asia Mixed Bugaeva, 2004, 2012; ValPaL
Haida Isolate North-America
Neutral Enrico, 2003
Lunda Bantu (Niger-Congo)
Africa Nom-Acc Kawasha, 2003
Koyraboro Senni Songhay Africa Neutral Heath, 1999
Lango Nilotic Africa Nom-acc Noonan, 1992
Mocoví Guaicuruan South-America
Mixed Juárez & Álvarez González, 2017
Tundra Nenets Uralic Europe Nom-Acc Nikolaeva, 2014
Tundra Yukaghir Yukaghir Asia Active Maslova, 2003
Neverver Austronesian (Central Vanuata)
Pacific Nom-Acc Barbour, 2012
Irarutu Austronesian (SHWNG)
13 4.1 Ainu
Ainu is an isolate language spoken on the Island of Hokkaido in Northern Japan. Ainu used to be spoken on other islands as well. There are three dialects of Ainu, namely Kurile (Islands), Taraika (Sakhalin) and Hokkaido. The latter dialect is the language that will be specifically discussed (Bugaeva, 2012: 461).
The language is spoken by the Ainus. According to Ethnologue, the language was spoken by 2 speakers in 2008, making the language practically extinct. The ethnic population, the Ainus, consists of 13.000 people (Ethnologue, 2019).
Bugaeva (2012: 461) describes Ainu as a polysynthetic, incorporating, and
agglutinating language, with a SOV constituent order and a mixed, but “basically tripartite”, alignment. The language is predominantly head-marking and prefixing.
Alignment. As said above, Ainu shows mixed alignment. Below, in examples 5-6, the alignment of Ainu is shown. The alignment for the first-person singular is
nominative/accusative with ku= marking the A/S (example 5a) and en= marking the O (example 5b). For the second person e= is used for singular and eci= for plural, and the third person is always zero-marked (example 5b). Tripartite alignment is found in the first person plural exclusive and in the “indefinite person”, there is distinct marking on for S, A and O (example 6). The intransitive predicates are indexed with S, and the transitive predicates with A and O.
(5a.) (káni) ku=mina 1SG 1SG.S=laugh ‘I laughed.’
(5b.) toan hekaci Ø=en=koyki that boy 3.A=1SG.O=bully ‘That boy bullied me.’
Bugaeva (2012: 473)
(6a.) neno é=iki yak a=e=kóyki na
like_this 2SG.S=do if IND.A=2SG.O=scold FIN ‘If you do that, you will be scolded [lit. one/they will scold you].’ (6b.) te ta rok=an ciki parka?
here LOC sit.PL=IND.S if be_good ‘May one sit here?’
14 Antipassive. Ainu has five devices for decreasing valency. One of these devices is the antipassive, traditionally referred to as the marker for a generalized object ‘(indefinite) person/thing’ in Ainu studies (Tamura, 1988, as cited in Bugaeva, 2012: 486). The antipassive is marked with the derivational prefix i-. The origins of this prefix can be found in the
pronominal indefinite O marker i-.
An example of the antipassive construction of Ainu is shown in example 7. The example shows that bivalent transitives become intransitives. The object in the antipassive construction of Ainu is completely blocked and the use of an oblique expression is not possible. The use of the antipassive prefix i- is shown in example 7b.
(7a.) ora-no nani usey ∅=∅=kar nea iwatarap then-ADV immediately hot_water 3.A=3.O=make that baby ∅=∅=huraye a ∅=∅=huraye a
3.A=3.O=wash ITR 3.A=3.O=wash ITR
‘Right away, my wife boiled some water, and washed that baby carefully.’
(7b.) ontaro or un ∅=i=huraye tub place ALL 3.S=APASS-wash
‘She did laundry in a tub.’ (lit. ‘washed things’)
Bugaeva (2012: 487)
The antipassive is less productive much less productive than applicatives or causatives in Ainu. It is only derived from certain verb classes such as
Perception/Cognition/Pursuit/Interaction and Affected Subject.
It is possible to apply the antipassive to trivalent transitives, although this is far less common than the application on bivalent transitives. This causes a valency decrease, not intransitivization, as in example 8b. Usually, the base trivalent verbs are semantic
ditransitives, as shown in examples 8a. These semantic ditransitive bases are verbs of transfer that contain a Theme (T) and a Recipient (R) in their role frame. The Theme
argument usually can be antipassivized, with the antipassivization of the Recipient argument being very rare cross-linguistically (Malchukov et al. 2010, as cited in Bugaeva 2012: 487).
15 However, in Ainu both the Theme argument and the Recipient argument can be made
antipassive.
(8a.) [pirka us-ke]T [∅=ona-ha]R ∅=∅=ko-puni
be_good place-POSS 3.A=father-POSS 3.A=3.O=to.APPL-raise/extend ‘(She) served the good parts of (meat) to her father.’
(8b.) [a=∅=kor ekasi]R hoski-no [nea okkaypo utar]A IND.A=3.O=have grandfather be_early-ADV that man PL Ko-iT-puni
to.APPL-ANTIP-raise/extend
‘The young men served my grandfather first.’
16 4.2 Haida
Haida is an isolate language spoken in Canada and Alaska. It has two dialects, Northern Haida and Southern Haida, both are discussed by Enrico (2003).
According to Ethnologue, the Northern dialect has two speakers and seven semi-speakers in Canada, with an ethnic population of 4550 people. The Northern dialect also has four speakers in Alaska and an ethnic population of 130 people. The Southern dialect has seven speakers and six semi-speakers in Canada and an ethnic population of 500 people.
Alignment. Haida displays neutral alignment with regards to case marking and agreement (Enrico, 2003: 46, 54, 84; Comrie, 2013a-b). This means that the S, A and the P are all marked in the same way.
Antipassive. In Haida the antipassive is accomplished by means of ta-incorporation. This incorporation’s function is to remove an unspecified complement and is often used to remove objects of transitive verbs. Ta loses its semantic content (Enrico, 2003: 86-87). The use of the indefinite object marker ta is shown below below in example 9b-c.
(9a.) xaw k’iina-7iid-ang tea be_hot-too-PRS ‘The tea is too hot.’ (9b.) ta k’iina-ang
ANTIP be_hot-PRS ‘The weather is hot.’
Enrico (2003: 87) (9c.) tl’@-gi nang ta dagihls, randll-gi dii gudang-ga
INDF-PP INDF ANTIP serve water-PP I want-PRS ‘You who are serving people, I want some water.’
Enrico (2003: 297)
Most underived subjectless verbs, including those that allow the antipassive, denote states or events of the “ambient environment”, such as the weather and tides (Enrico, 2003: 86). It is possible to apply the ta-incorporation to intransitive verbs, if the verb can refer to as a state or event of the ambient environment. This leads to a verb without complements (Enrico, 2003: 87).
The base of a ta-form is either an intransitive verb denoting a state of ambient environment, where the S-argument is lost, or a transitive verb, where the P-argument is
17 lost. The antipassive in Haida is lexically restricted. Many verbs where it is reasonable to expect a ta-form to exist, do not have one. The gaps for the ta-form are “numerous” (Enrico, 2003: 1265).
18 4.3 Lunda
Lunda, also known as Cilunda, is a member of the Bantu (Niger-Congo) language family. It is spoken in Zambia, Angola, and the Democratic Republic of Congo.
Lunda has a combined number of 403,000 speakers, in 2010 (Ethnologue). It has 225,000 speakers in Zambia (in 2010), and 178,000 speakers in Angola (in 2001). The number of speakers in the Democratic Republic of Congo is not mentioned. Due to the three
countries where Lunda is spoken, having different languages of administration and media of instruction, the exact number of speakers is not readily available (Kawasha, 2003: 5).
Lunda has multiple dialects of which are Koza, Ndembu, and Lunda of Ishindi are the major ones. The specific language described in Kawasha (2003: 4-5) is Koza, spoken in the Mwinilunga district.
Alignment. Lunda has a nominative-accusative alignment system (Bostoen et al., 2015; Kawasha, 2003: 243-244, 249).
Antipassive. The antipassive in Lunda is expressed with the suffix -an- on a limited number of verbs. When used, the verb takes an agentive A-argument and the P-argument is disallowed (Kawasha, 2003: 176). The suffix primarily functions to express what an agent frequently does or what is can possibly do. Bostoen et al. (2015: 743) refer to -an- as a dedicated antipassive marker, since reciprocity is expressed productively by the reflexive prefix -di- or residually by the extension -añan-. No example sentences of the use of the suffix -an- were given.
The antipassive in Lunda is not productive. The antipassive marker -an- can be used on a very limited number of verbs (Kawasha, 2013: 176). Kawasha does not elaborate on what type of verbs (dis)allow the antipassive, however a list of some of the verbs that can take the antipassive was given (see Appendix).
19 4.4 Koyraboro Senni
Koyraboro Senni is a language spoken on Mali, along the Niger river. It is a member of the Songhay language family. Heath (1999: 2) refers to Koyraboro Senni as a cosmopolitan language. It was spoken by 430.000 people in 2007. Of these speakers, 300.000 are monolingual. The ethnic population consists of 850.000 people (Ethnologue).
Alignment. The alignment system of Koyraboro Senni is neutral, both with regards to case marking and agreement (Heath, 1999: 157-163, 212, 299; Comrie, 2013a-b).
Antipassive. The antipassive construction of Koyraboro Senni is referred to by Heath (1999: 166) as the unspecified-object (UnspecO) derivation, but here it will be referred to as antipassive. In Koyraboro Senni, an -a suffix is combined with a transitive verb producing an intransitive verb. The transitive’s object is omitted, but the agent is kept. The -a suffix is homophonous with the mediopassive -a suffix. The mediopassive also has an alternative suffix, -andi.
The use of the a- suffix can have a mediopassive or an antipassive reading for many transitive verbs, however the surface subject of the mediopassive is often inanimate. In contrast, the surface S-argument of the antipassive are predominantly human. Heath lists four verbs that allow the antipassive, shown in table 2. These are not all the verbs that allow the antipassive in Koyraboro Senni (Heath, 1999: 166-167).
In example 10 the use of two of the verbs in table 2 in the antipassive construction is shown.
(10.) I ga nee ŋgey mma nees-a 3PL.SBJ IPFV say logo.PL.SBJ ST.IPFV measure-ANTIP tak-aa woo ha kaa ha kar-a
way-DEF.SG DEM INF come INF hit-ANTIP
‘They were figuring they would take aim in this way and then hit (the targets)’
Heath (1999: 167) Table 2. Verbs mentioned by Heath (1999: 166-167), as allowing the antipassive.
Transitive verb Antipassive meaning
Hit Do the hitting, make contact
Peel Do some peeling
Measure Take a measure, take aim
Sweep Sweep up, do the sweeping
20 There are limitations on the productivity of the antipassive. There are some
underived transitive verbs that avoid the -a suffix, instead these verbs use zero derivation for the antipassive. These verbs are not specifically listed by Heath. It is also uncommon for CVV, CVy, and CVw stems to have an -a suffix. It is not possible to combine the antipassive (and mediopassive) -a with the factitative-causative -andi (Heath, 1999: 166-167).
21 4.5 Lango (Uganda)
Lango is a (western) Nilotic language, spoken entirely in Uganda by the Langi. It is distinct from two other languages sometimes referred to as ‘Lango’, one is an eastern Nilotic
language, the other a Surmic language. The latter language is more commonly referred to as Didinga (Noonan 1992: 1).
According to Ethnologue, the user population consisted of 2,130,000 people in 2014 and is increasing. This number comes from a census based on tribal affiliation. It is a vastly higher amount than the estimates given in Noonan (1992: 1), which range from 300,000 in 1975 to 823,200 in 1988.
Alignment. Lango follows a neutral alignment pattern with regard to case marking. This can be seen in example 11 below, were there the S, the A and the P are marked in the same way. The verbal person marking of Lango follows an accusative pattern, as the A and the S are marked in the same way, whereas the P is marked differently (Noonan, 1992: 91-92).
Antipassive. In Lango, a verb root often has two, rarely three, stems that represent generalized valency schemes. These schemes are the Transitive, Activity Naming
(antipassive), and Secondary Argument.
The antipassive is referred to by Noonan as Activity Naming. The Activity Naming stem expresses a potentially transitive notion, but it does not reference a Direct Object. The A’s participation in an activity is solely referred to by the Activity Naming stem. The stem does not refer to activity towards any particular P-argument (Noonan, 1992: 125).
It is rare for a verb root to have stems of all three schemes. There is only one verb can clearly represent all three, namely nɛnɲ ‘see’. It is common for a two place predicate to only have a Transitive scheme, or to contrast the Transitive scheme with an Activity Naming Scheme or a Secondary Argument Scheme.
Below an example of a transitive construction, 11a, an antipassive (Activity Naming) construction, 11b, and the Secondary Argument, 11c, are shown. The morphemes were not separated in the original examples.
22 (11a.) lócə̀ ònɛ̀nò atîn
man 3SG-see-PFV child ‘the man saw the child’
(11b.) dákô bínô nénô woman 3SG-come-HAB see-INF ‘the woman will see’
(11c.) àtînn ànên
child 3SG-be_visible-PROG ‘the child is visible’
Noonan (1992: 126) The differences between forms can be seen in the verb roots. The Lango verb root has the basic form of C1VC2. In the infinitive, the transitive has a geminate C2, a final -o, and low tones on the root and final vowels. The antipassive form and the Secondary Argument form do not have a geminate C2, may have a final -o, and have various tonal contours (Noonan, 1992: 127-128). The changes in structure can be seen as a dedicated antipassive marker.
The productivity of the antipassive in Lango is limited. The transitive verb allows an antipassive construction, if the A-argument of the transitive verb is necessarily animate and the direct object necessarily inanimate, and if the verbs describes activities related to day-to-day life such as eating, physiological functions such as vomiting, economic activities such as planting, and business activities like trading (Noonan, 1992: 130).
23 4.6 Mocoví
Mocoví is a member of the southern branch of the Guaicuruan language family. It is spoken in Argentina. The language discussed by Juárez and Álvarez-González (2017) is spoken in Colonia Aborigen, in the center of the Chaco province.
According to Ethnologue, Mocoví had 2,780 speakers in 2012 with an ethnic population of 15,800. The largest group of speakers is found in the south of the Chaco province (Juárez and Álvarez-González, 2017: 228).
Alignment. Mocoví has no case marking for core arguments that are expressed by nominal phrases or independent pronouns (Juárez and Álvarez-González, 2017: 233). This shows a neutral alignment for case marking.
Mocoví has a split alignment for the indexing system. The language follows a nominative-accusative alignment for speech act participants (SAPs), and a tripartite
alignment for non-speech act participants. The first and second person predominantly show nominative-accusative alignment, but the third person mostly shows tripartite alignment (Juárez and Álvarez-González, 2017: 236-237). Due to the different alignment system, Mocoví is referred to as mixed in alignment in the Result chapter.
Antipassive. Mocoví has two antipassive markers, the suffixes -(ɑ)ɢɑn and -(ɑ)tɑɢɑn. There are some differences between the antipassive construction of these markers.
The -(ɑ)ɢɑn suffix highlights the activity denoted by the verb and is performed by the single active participant. The use of the suffix is shown in 11b. The P-argument of the
transitive clause (‘man’) has been deleted in the antipassive and the verbal index has changes from i- in 12a to -ɾ in 12b. It is possible for the P-argument to still be present in the antipassive construction, as shown in 12c, -(ɑ)ɢɑn still implies an antipassive construction as the third person index for S, -ɾ, is used.
(12a.) so pyoq i-tɑ-tɑk so yɑle CLF dog 3-sniff-PROG CLF man ‘The dog is sniffing the man.’
(12b.) so pyoq ɾe-tɑ-ɢɑn CLF dog 3-sniff-ANTIP ‘The dog sniffs.’
24 (12c.) so-mɑɢɑɾe-pi ɾ-owɑɢɑn-ɑɢɑn n-qopɑɢ
CLF-3-PL 3-hit-ANTIP INDF_POSS-stick
‘They cut firewood.’
Juárez & Álvarez-González (2017: 241)
The antipassive suffix -(ɑ)tɑɢɑn differs from -(ɑ)ɢɑn in two ways. One is that -(ɑ)tɑɢɑn can only be applied to a few Mocoví verbs, whereas -(ɑ)ɢɑn has an “extended occurrence” (Juárez and Álvarez-González, 2017: 247). This implies that the -(ɑ)tɑɢɑn suffix is not productive and that the -(ɑ)ɢɑn is productive.
The second difference is that the P-argument is always deleted when -(ɑ)tɑɢɑn is used to mark the antipassive construction. In 13b the use of -(ɑ)tɑɢɑn in the antipassive construction is shown. In 13a, the first person is encoded with ñ-, in the antipassive
construction (13b) the prefix s- is used. In 13b, it is shown that the antipassive construction implies a habitual meaning of the verb, rather than the effect as in 13a. The meaning of the abbreviations ATTR and MID were nog given.
(13a.) ɑyim ñ-ɑtɑɾen so ɾa-lola-ɢa-y-k
1SG 1MID-cure CLF 3-be_sick-NMLZ-ATTR-M ‘I cured the sick person.’
(13b.) ayim s-ataɾen-ataɢan 1SG 1-cure-ANTIP
‘I am a doctor.’ (Lit: ‘I cure.’)
25 4.7 Tundra Nenets
Tundra Nenets is a Uralic language spoken in European Russia and North-Western Siberia. In 2012, the ethnic population consisted of 44,600 people of which 21,900 spoke Tundra Nenets (Ethnologue).
Three large dialectal groups are often recognised, namely Western, Central, and Eastern/Siberian. There is relatively little dialectal diversity, the speakers of the different varieties can easily understand each other (Nikolaeva, 2014: 4).
Forest Nenets is the closest linguistic relative of Tundra Nenets. The languages were considered dialects of the single Nenets language for a long time, however there are
significant linguistic differences and an almost impossible mutual comprehension. Now, both languages are more often considered separate, but closely related languages.
Alignment. Tundra Nenets displays nominative-accusative alignment in case marking and agreement (Comrie, 2013a-b).
Antipassive. Based on the description of Nikolaeva (2014) and limited amount of examples, the antipassive construction does not seem to be very productive in Tundra Nenets. Tundra Nenets has a multiple derivational affixes that cause detransitivization when attached to a verbal stem. The suffixes -ŋko/-nc'o or -ŋkur- function as the antipassive markers. -nc'o is typical of Western dialects. Five verbs were given as examples of these suffixes, shown below in 14.
(14). tola- ‘to read’ tola-ŋko-/nc’o- ‘to do reading’ yera- ‘to guard, to keep’ yera-ŋko- ‘to be busy guarding’ p’ir’e- ‘to cook’ p’ir’e-nc’o-/ŋko- ‘to do cooking’ n’ada- ‘to help’ n’ada-ŋko-/nc’o- ‘to be helping’
xəl
̊
ta- ‘to wash’ xəl̊
ta-ŋku- ‘to do washingNikolaeva (2014: 226) When these antipassive suffixes are attached to the verbal stem in Eastern varieties of Tundra Nenets, the resulting verbs are one-place, meaning that the P-argument is not allowed to appear in the oblique phrase. In Western dialects the suffixes can be combined with the plural prolative objects (Nikolaeva, 2014: 226). This is shown in 15 below.
(15.) kniga-qm°na tola-nc’o-rka book-PL.PROL read-ANTIP-COMP ‘He reads books from time to time.’
26 Nikolaeva (2014: 226) 4.8 Tundra Yukaghir
Tundra, or Northern, Yukaghir is a member of the Yukaghir language family. It is related to, but distinct from Southern Yukaghir (Maslova, 2003: 1).
In 1987, there were a total of 170 Yukaghirs, of which ninety spoke Tundra Yukaghir (Maslova, 2003: 2). In 2010, there were 370 speakers of Tundra Yukaghir and Southern Yukaghir combined, with a combined ethnic population of 1,600 people (Ethnologue).
Alignment. Tundra Yukaghir displays an active alignment system (Heaton, 2017: 519). It is possible for the S-argument to be encoded in the same way as the A-argument, and sometimes the P-argument, depending on semantic factors (Maslove, 2003:10).
Antipassive. Only a small group of intransitive verbs exist that are derived from transitive verbs and that signify the same action as the transitive, but with an absent P-argument. Maslova (2003: 17) gives two examples, shown below in 16. There are multiple antipassive suffixes, namely -de-, -did’-, -din-, -d’e-, -die- and -d-. It is not clear how much of this
variation is phonological (Heaton, 2017: 462).
(16a) leŋ-de- eat-ANTIP-
‘be eating, have a meal’ (16b) pan-did’-
cook-ANTIP- ‘be cooking
Maslova (2003: 17) The antipassive in Tundra Yukaghir is not productive, as there are only a few verbs that allow the antipassive construction. Maslova does not discuss the qualities of the verbs that exceptionally allow the antipassive construction.
27 4.9 Neverver
Neverver is a member of the Austronesian language family, specifically of the Central Vanuatu branch. It is spoken on the Malekula Island in Vanuatu. It is the language of the Mindu and Sakhan people, and is spoken by approximately 560 people (Barbour, 2012: 1). This is less than half of the user population in of 1,250 people as mentioned on Ethnologue.
The modern version of Neverver is said to be based on the Mindu dialect. Despite some members of the Neverver community identifying as being of Sakhan descent, potential historical dialectal differences have been lost due to migration and regrouping of the
speakers.
Alignment. Barbour (2012: 264) writes that the primary subject is marked by a
combination of word order and verb agreement, the primary object is marked solely by word order, and the secondary object is marked by verbal morphology and word order. Neverver follows nominative-accusative alignment. The S and the A argument are placed before the verb and agree with the subject/mood prefix that is attached to the verb. The P-argument is placed behind the verb and the primary P-argument of a derived transitive verb is signalled by the applicative suffix -ikh on the intransitive verb stem (Barbour, 2012: 264-265).
Antipassive. In Neverver, reduplication has many functions, including
reflexive/reciprocal and durative. One of the functions of reduplication is the antipassive. Through reduplication, the P-argument is suppressed. In Neverver, unspecified object deletion and inherent object deletion can be distinguished.
In unspecified object deletion, the suppressed P-argument is considered irrelevant to the discourse and overt expression is therefore not necessary. In inherent object deletion, the suppressed P-argument can only be of a single referent type. The overt expression of the object argument is not allowed and unnecessary (Barbour, 2012: 180). Below some
examples of verbs that allow the antipassive construction are shown. (17a.) vus ‘carry something’
vusvus ‘carry a load’ (17b.) rukh ‘pick up something’
rukhrukh ‘gather coconuts’
Barbour (2012: 181) The antipassive in Neverver is not productive. Many transitive verbs do not allow valency change (Barbour, 2012: 179).
28 4.10 Irarutu
Irarutu is an Austronesian language, specifically of the South Halmahera-West New Guinea (SHWNG) branch. It’s spoken in West Papua on the island of New Guinea.
The Irarutu ethnic group consists of approximately 10,000 people. According to Jackson, it is not realistic to give a precise count of Irarutu speakers. Ethnologue gives an estimate of 4,000 speakers in 1987, Voorhoeve (1989; as cited in Jackson, 2014:7) estimates 5,000-6,000 speakers.
The specific dialect of Irarutu discussed in Jackson (2014) is the Fruata dialecy. Multiple dialects exist, however the exact amount is not yet clear. There is enough evidence to identify three dialects, but others such as Voorhoeve (1989) argue that there are seven (Jackson, 2014: 8).
Alignment. There is no case marking in Irarutu. Grammatical relations are established by word order and agreement. In a single argument clause, the S usually precedes the predicate. In a construction with a second and third person referent, the verb is often inflected to agree with the person of A. In a clause with two arguments, the A precedes the predicate and the O follows it. This leads to the conclusion that Irarutu follows a nominative-accusative alignment pattern with regard to word order and agreement (Jackson, 2014: 165).
Antipassive. In Irarutu, the presence of the -fe morpheme prevents the occurrence of an overt object. According to Jackson (2014: 107), the morpheme decreases valency and indicates a non-specified object. Below in 18a, the transitive construction is shown. In 18b, through the use of -fe the transitivity of the clause is lowered by deleting the overt object, instead a morpheme is used on the verb that indicates a non-specific ‘eat-things’ (Jackson, 2014: 107).
(18a.) it-ga uce 1Pn-eat papeda ‘we eat papeda’ (18b.) it-ga-fe
1Pn-eat-dtr ‘we eat (food)’
29 The -fe morpheme cannot be used with human P-arguments. This causes its
30 4.11 Collective Results
In table 3, an overview is given of the verbs that allow the antipassive, that were mentioned in multiple grammars. The verbs are listed in order of frequency.
In total, 20 verbs were mentioned as allowing the antipassive in more than one language. Some of these verbs are not part of the hierarchy. These verbs are ‘give birth’, ‘be angry’, ‘serve’, ‘sweep’ and ‘read’. ‘Speak’ also is not part of the hierarchy, but ‘talk’ is, which is the alternative translation given in both Irarutu and Lango.
Most verbs, 15 out of 20, were mentioned in two grammars. A top five of verbs in terms of frequency can be made (see table 3 and 4). ‘Cook’ is the most frequent verb appearing in five grammars as allowing the antipassive. ‘Eat’ appears four times, making it the second most frequent. This is followed by three verbs, ‘see’, ‘give birth’, and ‘hit’, these verbs were mentioned for three languages as allowing the antipassive.
Table 3. Overview of verbs that allow the antipassive, mentioned in the grammars of multiple languages
Verbs Ainu Lunda Koyraboro Senni
Mocoví Tundra Nenets
Tundra Yukaghir
Neverver Irarutu Lango Haida Number
1 cook cook Cook cook cook cook 5
2 eat eat eat eat eat 4
3 see see see see 3
4 give birth give birth (lit. be in pain) give birth give birth 3
5 hit hit hit hit 3
6 be angry be angry be angry 2
7 build build build 2
8 cut cut cut 2
9 grind grind grind 2
10 hear hear hear 2
11 help help help 2
12 name name name 2
13 peel peel peel 2
14 serve serve serve 2
15 shave shave shave 2
16 speak speak speak 2
17 sweep sweep sweep 2
18 teach teach teach 2
19 wash wash wash 2
31 5. Discussion
5.1 Main Findings
Table 4. Top five verbs
Verbs Number 1 cook 5 2 eat 4 3 see 3 4 give birth 3 5 hit 3
The aim of this study was to compare productive antipassive constructions in non-ergative languages. Specifically, to compare the restrictions placed on verbs that allow the antipassive, between languages and with Say (Forthcoming).
In table 4, the top five verbs are shown. This top five is not surprising based on the hierarchy and the five semantic properties of the ‘natural antipassive’. Four of the five verbs, excluding ‘give birth’, appear high in the hierarchy, in the first three levels. The ordering of these four verbs within the hierarchy is ‘eat’ > ‘see’, ‘hit’ > ‘cook’. The five properties also apply to these verbs.
‘Give birth’ is an outlier in this top five. In the definition of the antipassive in volume (Witzlack-Makarevich & Janic, Forthcoming) that Say is a part of, the lexical meaning of the verb in the antipassive should be identical in the transitive construction and the antipassive construction (Say, Forthcoming: 34). If this criteria is followed, ‘give birth’ in Ainu cannot be included in the top five, as the literal meaning is ‘to be in pain at’, shown in 19 below. However, verbs like these can be referred to as “lexicalized antipassive”.
(19.) Ikoni. i-koni
APASS-be.in.pain.at
‘Give birth / she gave birth.’
Bugaeva, Anna. 2013. Ainu Valency Pattern. Example 186
Of the 15 verbs that are mentioned for two languages, 11 are a part of the hierarchy. The hierarchy consists of 14 levels. Most of the verbs are high in the hierarchy, specifically
32 ‘name’, ‘teach’ and ‘grind’ are in the middle range of the hierarchy, ranging from the sixth to tenth level. ‘Build’, ‘peel’ (level 11) and ‘talk’ (level 12) are at the lower end of the hierarchy.
The four verbs that are not part of the hierarchy are ‘be angry’, ‘serve’, ‘sweep’ and ‘read’. ‘Serve’, ‘sweep’ and ‘read’ would appear high on the hierarchy as the semantic properties do apply well to these verbs. ‘Be angry’ would fit most at the low end of the hierarchy, similar to ‘be sad’, as the semantic properties of the natural antipassive do not apply well to this verb.
‘Be angry’ appears to be an intransitive verb in Ainu. However, the Ainu verb has a cause of the anger inherent in the verb meaning, as shown in example 20a. Example 20b shown the use of the antipassive marker i-, causing the ‘be angry’ meaning. Whether the same applies to Lunda is unknown, as no examples were given in Kawasha (2003), it is also possible that there is a special patientive S.
(20a.) ruska
‘be angry with/because of sth’ (20b.) i-ruska
Antip-be.angry.because.of ‘be angry’
33 5.2 Limitations and future
When considering the results of this study, a number of factors should be taken into account. First, this study was based on the information included in the grammars of the languages. Due to this, there were differences in the amount of verbs found that allow the antipassive construction. This ranges from three verbs (Haida) to twenty-three verbs (Ainu), with Lango being a bit of an outlier with seventy verbs mentioned (see Appendix). The grammars also varied in descriptions of the limitations placed on the antipassive
construction. It is very likely that there are more verbs that allow the antipassive, that were not mentioned in the grammars. There was also no negative evidence, as all languages had non-productive antipassives. How the properties compare to productive antipassives cannot be concluded from this study. A more systematic approach, for example through elicitation and the use of other sources on the languages, such as (spoken) language corpora, could lead to more comprehensive lists of ‘antipassive verbs’, and more evidence for or against the five properties and the hierarchy.
Another factor to consider, is the potential difference in defining “productivity” from Polinsky (2013). Polinsky does not define her criteria for productivity. Due to this, it is possible that languages that have been defined as, for example, partially productive by Polinsky, have been identified as non-productive here. This is also possible as there has been no distinction made between “not productive” and “partially productive”, which Polinsky has done. Instead, the antipassive was either productive or non-productive.
34
5 Conclusion
This paper tried to compare the verbs that allow the antipassive construction to the semantic properties of the ‘natural antipassive’ and to the hierarchy of 80 verbs from the ValPaL project, based on their likelihood the enter the antipassive.
The majority of the twenty verbs found to allow the antipassive construction in two or more languages of my sample, appeared on the high or middle range of the hierarchy. These verbs contained all or some of the semantic properties as mentioned by Say
(forthcoming), corresponding to their position on the hierarchy. A top five of verbs in terms of frequency of appearance could be made. Four of these verbs appear high, levels 1 to 3, on the hierarchy and satisfy the semantic properties.
The restrictions of the antipassive constructions and the verbs that (dis-)allow it, is an under-examined topic. More systematic research can provide more knowledge of the
35 References
Ainu example No. 186. In: Bugaeva, Anna. (2013). Ainu Valency Patterns. In: Hartmann, Iren & Haspelmath, Martin & Taylor, Bradley (eds.) 2013. Valency Patterns Leipzig. Leipzig: Max Planck Institute for Evolutionary Anthropology. (Available online at
http://ValPaL.info/languages/ainu/examples/186, Accessed on 2020-06-17) Aldridge, Edith. (2007). Minimalist analysis of ergativity. Sophia Linguistica 55. 123–142.
Bostoen, Koen, Dom, Sebastian & Segerer, Guillaume. (2015). The Antipassive in Bantu. Linguistics. 53(4), 731-772.
Bugaeva, Anna. (2012). Southern Hokkaido Ainu. In Tranter, N. (ed.), The languages of Japan and
Korea. (Routledge Language Family Series). London: Routledge, 461-509.
Cooreman, Ann. A functional typology of antipassives. In Fox, B. & Hopper, P. (eds.), Voice: Form and
Function. Amsterdam: John Benjamins, 49-88.
Comrie, Bernard. (2013a). Alignment of Case Marking of Full Noun Phrases.
In: Dryer, Matthew S. & Haspelmath, Martin (eds.) The World Atlas of Language Structures Online. Leipzig: Max Planck Institute for Evolutionary Anthropology. (Available online at http://wals.info/chapter/98, Accessed on 2020-06-22.)
Comrie, Bernard. (2013b). Alignment of Case Marking of Full Noun Phrases. In: Dryer, Matthew S. & Haspelmath, Martin (eds.) The World Atlas of Language Structures Online. Leipzig: Max Planck Institute for Evolutionary Anthropology. (Available online at
http://wals.info/chapter/98, Accessed on 2020-06-22.)
Enrico, John. (2003). Haida Syntax. (Studies in the anthropology of North American Indians). Lincoln: University of Nebraska Press.
Eberhard, David M., Gary F. Simons, and Charles D. Fennig (eds.). 2020. Ethnologue: Languages of the
World. Twenty-third edition. Dallas, Texas: SIL International. Online version:
http://www.ethnologue.com.
Hartmann, Iren & Haspelmath, Martin & Taylor, Bradley (eds.) 2013. Valency Patterns Leipzig. Leipzig: Max Planck Institute for Evolutionary Anthropology. (Available online at http://ValPaL.info, Accessed on 2020-06-22)
Heath, Jeffrey. (1999). A Grammar of Koyraboro (Koroboro) Senni- The Songhay of Gao, Mali. (Westafrikanische Studien, book 19). Rüdiger Köppe.
Heaton, Raina. (2017). A typology of antipassives, with special reference to Mayan. PhD Thesis, University of Hawaii, Manoa.
Heaton, Raina. (2020). Antipassives in Crosslinguistic Perspective. Annual Review of Linguistics. 6, 131-153.
Jackson, Jason. (2014). A Grammar of Irarutu, a language of West Papua, Indonesia, with historical
36 Janic, Katarzyna. 2016. L’antipassif dans les langues accusatives. Brussels: Peter Lang.
Janic, Katarzyna & Witzlack-Makarevic. (Fortcoming). The multifaceted nature of the antipassive construction. In Katarzyna Janic & Alena Witzlack-Makarevich (eds.), Antipassive
constructions in the languages of the world. Amsterdam: John Benjamins.
Juárez, Christian & Álvarez-González, A. (2016.) The antipassive marking in Mocoví. Forms and functions. In Albert Álvarez-González & Ía Navarro (eds.), Verb Valency Changes: Theoretical
and typological perspective. Amsterdam: John Benjamins.
Kawasha, Boniface. (2003). Lunda Grammar: A Morphosyntactic and Semantic Analysis. Doctoral dissertation, University of Oregon.
Kulikov, Leonid. (2010). Voice Typology. In JJ Song (ed.), The Oxford Handbook of Linguistic Typology, 368-398. Oxford: Oxford University Press.
Malchukov, Andrej. (2005). Case pattern splits, verb types, and construction competition. In
Mengistu Amberber & Helen de Hoop (eds), Competition and Variation in Natural Languages, 73–117. London: Elsevier.
Maslova, Elena. (2003). Tundra Yukaghir. (Languages of the World/Materials, 372.) München: Lincom. viii+97pp.
Nikolaeva I. (2014). A Grammar of Tundra Nenets. Berlin: Mouton de Gruyter
Noonan, Michael. (1992). A Grammar of Lango. (Mouton Grammar Library, 7.) Berlin, New York: Berlin: Mouton de Gruyter. xvi+352pp.
Palmer, F. (1994). Grammatical Roles and Relations. Cambridge, UK: Cambridge Univ. Press
Polinsky, Maria. (2013). Antipassive Constructions. In: Dryer, Matthew S. & Haspelmath, Martin (eds.)
The World Atlas of Language Structures Online. Leipzig: Max Planck Institute for Evolutionary
Anthropology. (Available online at http://wals.info/chapter/108, Accessed on 2020-06-17.) Polinsky, Maria. (2017). Antipassive. In The Oxford Handbook of Ergativity, ed. J Coon, D Massam, L
Travis, pp. 308–31. Oxford, UK: Oxford Univ. Press
Say, S. (Forthcoming). Antipassive and the lexical meaning of verbs. In Katarzyna Janic & Alena Witzlack-Makarevich (eds.), Antipassive constructions in the languages of the world. Amsterdam: John Benjamins.
Silverstein, Michael. (1972). Chinook Jargon: Language Contact and the Problem of Multi-Level Generative Systems, I. Language, 48(2), 378-406.
Silverstein, Michael.( 1976). Hierarchy of Features and Ergativity. In Dixon, R. M. W. (ed.),
Grammatical Categories in Australian Languages, 112-171. Canberra: Australian National
University.
Song, JJ. (2001). Linguistic Typology: Morphology and Syntax. Harlow, UK: Pearson Educ.
Vigus, M. (2018). Antipassive constructions: Correlations of form and function across languages. Linguistic Typology, 22(3), 339-384.
37 Appendix
Ainu
Verb Eat Shave Give Think Wash Wipe Search for Hear Tell Hug Help Name Load Teach Fear Carry Grind Like Leave MeetFeel pain – meaning “give birth” in antipassive (i-koni)
Raise/extend to (vert. serve) Be angry
Haida
Verb Throw (rocks) Serve Be.hot38
Lunda
Transitive Antipassive Meaning
Die Be deadly, be dangerous
Deceive Be false, deceitful
Bite Be biting, likely to bite
Insult Be insulting
Curse Be one who curses
Be angry Be quarrelsome
Be drunk Be intoxicating
Hear Be audible, be famous
Appear Appear, be visible
Name Be famous See Be visible Be able to Be possible Verb Die Deceive Bite Insult Curse Be angry Be drunk Hear Appear Name See Be able to
Koyraboro Senni
Transitive Antipassive Meaning
Hit Do the hitting, make contact
Peel Do some peeling
Measure Take a measure, take aim
Sweep Sweep up, do the sweeping
39
Lango
Transitive Antipassive Meaning; when empty the meaning is the same as the transitive
Eat See Play Run Chew To be sufficient p. 88 To pound To pound To bore To be tired of To accumulate To accumulate To unshell To unshell To collect, gather To roast, fry To eat
To sell To engage in business
To write, sow To write, sow
To thresh
To collect gather (ander ww) To break up (fight) To build To drive in (nail) To beat To kick To sweep up To speak To blow
To weep over To cry
To dig
To blow on To blow
To gather, harvest To pick out among To steal To read, recite To sift To beg To dream To pray to To pray
To rejoice over To rejoice
To defeat, exceed To rule
To slash, clear To call To shave To drink To dance To tremble To see To chew
40
To marry To get married
To laugh at To laugh
To give birth to To give birth To vomit
To run from To run
To collect (vegetables) To peel
To hide from To hide
To plant To winnow To teach To grind
To be insufficient for To be insufficient
To run from To run
To winnow To cook To ladle out To play To try To forge To work, do To work To tie, bind To sweep To sieve To climb
Mocoví
Transitive Antipassive Meaning
Agan:
Cut Take
Hit Hit (at)
Kill Sniff Pile
Atagan:
Cure (I am curing) Am a doctor (let. I cure) Know (vert. Think)
Tundra Nenets
Transitive Antipassive Meaning
To read To do reading
To guard, to keep To be busy guarding
To cook To do cooking
To help To be helping
41
Yukaghir Tundra
Eat Cook Go Cross (p. 82)Neverver
Cut See CookCarry (of fire) Rake
Buy/go shopping Cast
Give birth Carry s.t.
To scoop out the insides of s.t. Clear ground
Pick up s.t.
Transitive Antipassive Meaning
tn tn - Roast/cook Titn- Do cooking
Gav Gav – rake Gavgav – do raking
Vul Vul – buy Vulvul – go shopping
Tuv (throw) Tuv – cast (of a round object) at s.t. Tuvtuv – cast (of a round object) at s.t. (unspecified)
Tuv-ikh
(throw)
Tuv-ikh – cast s.t. at s.t. Tuvtuv-ikh – cast s.t. at s.t. (unspecific)
Khit Khit – see s.t. Khitkhit – to look
Vus Vus – carry s.t. Vusvus – carry a load
Leb Carry s.t. Lebleb – carry a load of garden
produce
Yel To scoop out the insides of s.t. Yelyel – shell out coconuts
Rakh Clear ground Rakhrakh – pull weeds
Rukh Pick up s.t. Rukhrukh – gather coconuts
Leb Give birth to s.o. Lebleb – give birth to lots of
babies/large litter
Irarutu
Transitive Antipassive Meaning
Eat Eat things
Stab/pierce Sew-things Cook Cook-things Shoot-things Speak To hit Smoke Work
