Katarzyna Janic & Alena Witzlack-Makarevich: Antipassive: typology, diachrony and related constructions

1 Introduction

Antipassive: Typology, diachrony, and related constructions is volume 130 in the John Benjamins series Typological Studies in Language. The editors Katarzyna Janic and Alena Witzlack-Makarevich present a selection of papers that were given at the workshop ‘The crosslinguistic diversity of antipassive: function, meaning, and structure’ which took place during the 49th Annual Meeting of the Societas Linguistica Europaea at the University of Naples Frederico II in 2016.

In edited collections of this kind, antipassives are often a subtopic in the treatment of ergative languages (e.g., Coon et al. 2017; Johns et al. 2006; Polinsky 2017a; Rill 2017). Janic and Witzlack-Makarevich’s volume is the first book-length treatment dedicated solely to antipassives.

While antipassives generate quite a lot of interest in both formal and functional literature, they are rarely the main topic of these types of multi-faceted edited volumes. This book frames the papers not just in comparison to or as an offshoot of transitive constructions but treats them in their own right, which is quite refreshing and highly appreciated.

As the editors point out in their immensely helpful introductory chapter, antipassives have been viewed through many and varied lenses. As is often the case, whether someone considers a particular construction an antipassive depends on their research focus, the languages to be examined, and the general theoretical or descriptive framework adopted. Therefore, formal criteria are outlined in the introductory chapter to limit the frame used to discuss antipassive constructions. The criteria are: (1) the same verb with the same number of participant roles is used in an equivalent transitive construction, (2) the A^[1] argument is the sole S argument in the antipassive, and (3) the P argument is either an oblique or unexpressed.

These criteria offer an excellent starting point for the discussions, mostly from a functional-typological perspective, of antipassives and antipassive-like constructions, that are included in the present volume.

The book provides a wide range of data concerning different languages, language families, and approaches to description and analysis. The function and form of the constructions are organized into four broad thematic sections. The first and longest part, with six contributions, addresses the event semantics of the antipassive. The second part covers antipassive marking on the verb in four contributions. The third part addresses the diachrony of antipassives in three contributions, and the fourth part addresses constructions that show very few, if any, of the properties of antipassives as defined in the introductory chapter in five contributions.

The section on terminology in the introductory chapter is quite useful, although the terminology is not used consistently in the volume. Especially the use of the term ‘transitivity’ seems to veer between a formal syntactic usage that covers transitive agreement morphology and a semantic usage that sometimes coincides with valency. Crucially, the labels for the arguments themselves, S, A, and P/O, are used consistently in this volume, which offsets the terminological confusion I contended with on occasion. It should be pointed out that the inconsistent use of the term ‘transitivity’ is quite common in general in the literature and, therefore, not surprising, albeit problematic if not initially defined in a chapter.

The points of variation in how antipassives may be realized in different languages are outlined as 1) the realization of the P argument, 2) antipassive marking on the verb, especially its relationship with reflexive/reciprocal markers, aspectual markers, light verbs, nominalizations, person markers, or benefactive/malfactive markers and 3) the semantic and discourse pragmatic functions of antipassives and antipassive-like constructions.

Backgrounding such as demotion to non-core arguments and loss of referentiality, indefiniteness, low individuation and suppression of P arguments are probably the most common topics discussed as properties of antipassives, especially in chapters 2, 4, 5, 7, 10, and 15. On the other hand, while topicalization and foregrounding of A arguments are common, they seem to draw less focus in discussions of antipassives in general and in this volume. Less bounded aspectual semantics are also common in antipassives, as discussed in parts 1 and 2 especially. The only syntactic difference between transitives and antipassives happens when the inaccessibility of the A argument to wh-movement, relativization, and coordination necessitates a change to the S argument. Languages differ regarding whether all these processes can access the A argument or only some (Polinsky 2017a). Very few examples of syntactic ergativity appear in this volume, which raises the question of why discussions of antipassives and deep ergativity are so rare.

The productivity of a valency changing process is a major aspect to investigate. If a construction is severely restricted to either just one or two verbs, or one type of discourse context, it certainly cannot be described as a major component of the grammar of a language. This volume provides some examples (such as in chapters 13 and 16) where the productivity of the antipassive is very low.

In what follows, I summarize individual chapters/contributions in Sections 2–5, before providing an overall evaluation of the volume in Section 6.

2 Part 1: Lexical semantics and event representation of antipassive constructions

Marianne Mithun ‘Antipassive propensities and alignment’ discusses antipassives in languages with agent/patient alignment. The main underpinning of these constructions is that they serve to background themes, goals, or patients that are not topics for one reason or another. The languages discussed are Lakota (ISO 639-3 lkt) (Siouan), the isolate Haida (ISO 639-3 hax/hdn), and Central Pomo (ISO 639-3 poo) (Pomoan), where the construction eliminates the P argument entirely. This property stands out since most other antipassives discussed in the volume have P optionally present. Mohawk (ISO 639-3 moh) (Iroquoian) and Hiligaynon (hili1240)^[2] discussed in the second part of the chapter demonstrate that the antipassive function of backgrounding the P argument can occur in both ergative and nominative accusative languages. Mithun addresses why such a large body of research argues that antipassives exist only in ergative languages. It seems that this is more a question of perspective and how one applies the label of antipassive. If we consider any backgrounding of P arguments an instance of antipassive, the alignment system is not important. However if antipassives are defined as detransitivization with a switch of an A argument to an S argument, then there is much more to be said about antipassives in ergative languages than in non-ergative languages. For instance, the difference between antipassives and actual split-ergativity is not always clear, and the properties that trigger splits or antipassives are often very similar (Spreng 2010, 2020). Moreover, the question of whether there is a change of status for the subject is not an issue with nominative-accusative alignment languages (Harris 1990; Nash 2017). However, I agree that the discussion of antipassives should not be limited to ergative alignment languages.

Raquel-María Sapién, Natalia Cáceres Arandia, Spike Gildea and Sérgio Meira ‘Antipassive in the Cariban family’ discuss data from five Cariban languages, namely Akawaio (ISO 639-3 ake), Hixkaryana (ISO 639-3 hix), Kari’nja (ISO 639-3 car), Tiriyó (ISO 639-3 tri), and Ye’kwana (ISO 639-3 mch). The presented examples are identified as antipassives and selected from extensive corpora containing different genres. In these languages, not all detransitivized constructions can be described as antipassives, as the authors show. Some of the constructions discussed serve to background arguments other than P, such as with reflexives, middles, anticausatives, and passives. The wide range of detransitivization strategies in service of backgrounding one argument provides an interesting view on the functions of detransitivization in general and not only in the Cariban languages. The chapter especially stands out for its detailed discussion of the advantages and disadvantages of large-scale data collection. I especially appreciated the note about the problems that may arise when one cannot return to the field to ask more in-depth questions.

Jessica Denniss ‘Aspect and modality in Pama-Nyungan antipassives’ provides insights into antipassives in the Pama-Nyungan family (pama1250) and explains why these constructions often correspond to unbounded verbal semantics and how that relates to the backgrounding of certain P arguments. The relationship between a lower degree of transitivity and unbounded aspect has long been known (Hopper and Thompson 1980), but this chapter presents a convincing explanation for why this connection exists. The author accounts for the imperfective aspectual readings and the increase in volitional modality of antipassives in these languages by proposing an aspectual marker that connects this modality with atelicity. Modality especially does not appear much in discussions of antipassives cross-linguistically, so I was surprised and intrigued to see a connection to transitivity here. I appreciated this well-argued contribution that outlines the reasons for antipassives focusing on A arguments and the volitional effects on event semantics in antipassives.

Claire Moyse-Faurie ‘Antipassive constructions in Oceanic languages’ discusses multiple constructions that have been labeled antipassives in the past (Cooreman 1994) and concludes that despite the low individuation of the P argument, the low transitivity, and the occasional habitual aspect interpretation, these rarely occurring properties do not merit the label antipassive. In addition, none of the constructions that are discussed in the chapter have an antipassive marker, demonstrating that there is very little formal, semantic, or morphological evidence to describe these as antipassives. Moreover, they barely fulfill the criteria outlined in the introduction chapter or even the larger set of properties in Polinsky (2011).

Sergey Say ‘Antipassive and the lexical meaning of verbs’ addresses the question of whether there are some verb types that are more easily antipassivized than others. The author draws verb type examples from multiple unrelated languages, making the point that their semantic and syntactic properties overlap enough to characterize them as forming ‘natural’ antipassives. It appears that what is considered ‘natural’ in this chapter are simply verbs with lexical semantics that allow a limited range of objects and are generally less marked when intransitive. This group includes verbs that most frequently only take cognate objects such as eat. Ultimately, the author argues that semantic properties of the P argument are responsible for the verb meanings and the wide variation of the properties of antipassives cross-linguistically. The cited proposal by Tsunoda (1981) has illustrated the correlation between higher markedness, i.e., requiring overt morphology for high-impact transitive verbs, and ‘natural’ antipassives for verbs that can be detransitivized without an antipassive marker. This correlation has been shown for various languages and in many works on aspect and detransitivization (Abraham 1996; Rothstein 2004; Tenny 1995) in much previous work.

Anna Bugaeva ‘Unspecified participant: A case of antipassive in Ainu (ISO 639-3 ain)’ illustrates the role of the antipassive marker as descended from the incorporation of a generic noun. This antipassive marker applies to verbs in the middle of a transitivity hierarchy (Hopper and Thompson 1980). The author argues convincingly that the Ainu indefinite person marker has become an antipassive marker with common effects on P arguments. The data are well presented, and the author provides extensive examples to support her argument.

3 Part 2: Antipassive marking

Katarzyna Janic ‘Variation in the verbal marking of antipassive constructions’ provides a comparison of the different types of coding on the verb commonly termed antipassive marker while also giving some reference to the synchronic and diachronic relations to other functions of the same markers. The chapter covers segmental morphemes, sound changes on the verb, and null marking. Languages from various families are discussed, including languages with all kinds of alignment patterns. While many of these antipassive marking strategies align with detransitivization, this chapter demonstrates that not all do. This chapter highlights why it may be difficult to label every detransitivization with P-backgrounding an antipassive if it is not always clear if an antipassive has some kind of overt marker, a null marker, or nothing. As I show in Spreng (2006) for Inuktitut, there are good reasons to suspect that not everything that looks like an antipassive has an actual antipassive marker or the antipassive marker might not even be responsible for detransitivization (Basilico 2019). Simple object omission or backgrounding is often considered an antipassive. At the same time, it appears to be simply a lexical property of certain types of verbs with an argument structure that easily allows the omission or backgrounding of a P argument (Levin 1993).

Denis Creissels ‘Antipassive derivation in Soninke (West Mande)’ illustrates antipassives in Soninke (ISO 639-3 snk) that have none of the semantic and discourse-related functions of antipassives in other languages. However, Soninke does have a dedicated antipassive marker and a detransitivizer with limited productivity. Both markers co-occur in a detransitivized construction where the A argument has become the S argument. Both constructions demonstrate morphological, possibly syntactic, and marked detransitivization overtly shown on the verb. Also interesting is in these constructions, P must be expressed in almost all cases, making the construction stand out based on general formal requirements of the language (P must be expressed) while still showing morphological detransitivization through the overt marker. The semantics of these constructions are quite interesting and somewhat unexpected. They have none of the unbounded or habitual aspectual interpretations, the P argument must be overt, while the constructions as such are highly productive.

Christian Juárez and Albert Alvarez Gonzalez ‘Explaining the antipassive-causative syncretism in Mocoví (ISO 639-3 moc)’ explain the unusual syncretism of the Mocoví antipassive marker with a causative marker by showing that the P arguments that are being backgrounded in causatives and regular antipassives are different. The data are interesting and support the argument quite convincingly. A comparison with applicatives would have been interesting to see if the backgrounding of the P argument would resemble antipassives or causatives.

Alejandra Vidal and Doris L. Payne ‘Polyfunctional vanka- in Nivačle and the antipassive category’ clearly outline the multiple functions of the antipassive marker in Nivačle (ISO 639-3 cag). It appears to be more of a general detransitivizer, especially since it also occurs in a middle construction. The first part of this chapter was well argued and supported by the data, while the diachronic section on the development of vanka- is more speculative in that there is very little evidence that could support the author’s view that the morpheme has developed from a combination of the third person marker va-, a ‘cislocative’ or middle marker n-, and an indirect possession marker ka-. The relationship between three morphemes to be combined to a detransitivizing antipassive marker appears logical given the semantics of the constructions in which vanka- occurs. However, the supporting evidence appears quite tenuous, which the authors acknowledge in their conclusion, emphasizing the need for further study.

4 Part 3: Diachrony of antipassive constructions

Sandra Auderset ‘The antipassive and its relationship to person markers’ discusses the diachronic development of person markers as antipassive markers. While there is a precedent of such a development for passive markers (Guillaume 2011; Siewierska 2010), the relationships that are being described appear tenuous and speculative. While it is commendable that the author considers a wide variety of language families, this is also a drawback. The pathways sketched for antipassive development are unlikely to be similar across language families. Particularly puzzling is how politeness was considered a pathway from regrammaticalizing third and first person markers as antipassives. Many questions are left unanswered in the chapter, and here I would like to pose a simple one. Why would antipassives be a regrammaticalization of first and third person but not second person? Crosslinguistically, first and second person tend to pattern together as discourse participants in both synchronic and diachronic relations. Thus further investigation may be needed to arrive at an explanation for this rather unusual patterning.

Guillaume Jacques ‘Antipassive derivations in Sino-Tibetan/Trans-Himalayan and their sources’ addresses the diachronic sources of antipassive constructions in Gyalrongic (rgya1241), Kiranti (kira 1253), West-Himalayish (tibe 1275), Old Chinese (ISO 639-3 och), and Dulong-Rawang (rawa1265, dulo1243). The sources proposed are incorporation of generic nouns, verbalization of action nominalizations, and reflexive/middle markers. All sources are well supported in the data from various languages that the author draws on. However, since the constructions appear to be a fairly recent development, their productivity is affected. Most of them are quite unproductive except for the ones in the languages where the antipassive developed from action nominalizations. The author emphasizes that the diversity of the sources of antipassives in Sino-Tibetan requires much more in-depth research, especially considering recent diachronic developments.

Doris L. Payne ‘The profile and development of Maa (Eastern Nilotic) antipassive’ discusses an intransitive construction in Maa (ISO 639-3 cma) that has somewhat tenuous connections to antipassives but obvious correlations to imperfective aspect when the P argument is not overt. The construction has an antipassive marker that appears to contribute to the aspectual semantics. The chapter discusses whether the marker originated directly from the verb give or whether it evolved due to drift influence from Southern Nilotic. This tentative exploration is interesting since such a correspondence has not been seen with antipassive markers as far as I am aware. It is also particularly surprising since there is a common assumption that light verbs have their origins in main verbs, even though there is little evidence to support this view (Bowern 2008).

5 Part 4: Fuzzy boundaries

Peter Arkadiev and Alexander Letuchiy ‘Indirect antipassives in Circassian’ discuss antipassives in the Circassian (circ1239) languages. The constructions are interesting in their effect on the P argument. In these languages, not every antipassive targets a direct object for elimination and overt P arguments are not just backgrounded but deleted, suggesting that the constructions that are discussed here may not be derived from transitive constructions (Polinsky 2016, 2017a, 2017b). Especially this last property suggests that P arguments are not demoted. The option to target different arguments for detransitivization requires a clear distinction between different types of objects in these languages, which does not seem relevant in other constructions discussed in this volume. They follow the general semantic verbal properties in that they tend to be atelic and occur with a non-specific P argument. Complement clauses of speech verbs also occur in the antipassive construction. Most data discussed in this volume and in general are concerned with main clauses, so this is something new. While antipassives are obligatory in relative clauses in some languages, complement clauses are not often discussed in this context (Dixon 1994; Rill 2017). More crosslinguistic research on antipassives in complement clauses is sorely needed.

Bernard Comrie, Diana Forker, Zaira Khalilova, Helma van den Berg ‘Antipassives in Nakh-Daghestanian languages: Exploring the margins of a construction’ address constructions that have very few, if any, of the properties outlined in the introduction chapter and they are mostly unproductive and rarely related to transitive constructions. One might ask if these are not simply aspectual markers removing telicity as proposed by Tatevosov (2011). The main property they have in common with antipassives is the shift to unbounded aspects such as durativity, atelicity, and iterativity. One argument in support of treating these constructions as antipassives is the still-present P argument. There is quite a lot of variation among the constructions discussed, and classifying them all under an antipassive umbrella seems somewhat forced. We find similar properties in partitive constructions in Finnish, but to my knowledge, such constructions have not been discussed within the context of antipassives (Huumo 2010; Kiparsky 2001; Kratzer 2004). This section, and especially this chapter, underscores how a wide-angle approach can create additional “fuzziness” in the treatment of a single topic.

Raina Heaton ‘Antipassive and antipassive-like constructions in Mayan languages’ follows the criteria outlined in the introduction chapter to evaluate several constructions from Mayan (maya1287) languages. In particular, Mayan languages have four types of detransitivized constructions: absolutive, incorporating, agent focus, and reflexives/reciprocals. The author finds that only the first follows most of the criteria for antipassives. Not even all types of absolutive constructions are found to be antipassives. However, those that are have overt verbal antipassive marking, coindexing of the agent on the verb, and an oblique or unexpressed patient. This chapter is a very clear exploration of the constructions in question and their properties.

Charlotte Hemmings ‘When an antipassive isn’t an antipassive anymore: the Actor Voice construction in Kelabit’ compares the properties of the Actor Voice construction in the Austronesian language Kelabit (ISO 639-3 kzi) to the properties of antipassives in West Greenlandic (ISO 639-3 kal). Using the specific properties of antipassives from an unrelated language as criteria for evaluation is an unusual choice since general criteria are provided at the beginning of the volume. Moreover, the multidimensional properties of antipassives make comparing antipassive properties of unrelated languages very difficult and it might appear as if comparing apples to oranges. Thus the multidimensional properties of antipassives make comparing completely unrelated languages seem ill-advised. What is interesting, however, is the comparison with Tagalog (also Austronesian), which appears to have the semantic and discourse-related properties of antipassives as outlined in the introduction while none of the syntactic properties. Kelabit, on the other hand, has most of the syntactic but few of the semantic properties of antipassives. This phenomenon may have implications for language change within the Austronesian language family. Similar changes can be found in Inuit languages; in these languages, there is a shift in some of the Eastern Canadian variants towards the antipassive being the more common construction while the transitive construction is losing ground (Johns 2006).

Fernando Zúñiga and Beatriz Fernández ‘Antipassivization in Basque revisited’ discusses three constructions in Basque (ISO 639-3 eus), the ergative displacement construction (Laka 1989), the ari-progressive (Laka 2006), and participial clauses (de Urbina and Uribe-Etxebarria 1991). All of them have, over time, been treated as possible instances of antipassives. Unfortunately, this paper comes to no new conclusions. It seems to reaffirm all previous work that concluded that Basque does not have any constructions that deserve the antipassive label, even under a fairly loose definition of antipassives. What is valuable, however, is that the authors do not simply consider the semantics of the constructions and decide that they are antipassives but also focus on whether the formal properties align with the semantic and discourse-related properties.

6 Evaluation

Edited volumes based on a thematic workshop or conference are often helpful in giving the reader a state-of-the-art overview of current work on a particular topic, and this volume accomplishes that spectacularly. It brings together primarily typological and functional approaches to antipassives under a clear umbrella that limits the applicability of the term antipassive within the volume. However, some chapters seem to pay less attention to those criteria without clearly stating why they would include the data they are discussing in such a volume. While it is highly appreciated to see new data that may and likely should be discussed in this context, this reader finds it difficult to be satisfied with purely semantic or pragmatic definitions of antipassives. A stricter application and reference to formal criteria in selecting or editing the chapters would have helped this reader evaluate and appreciate the wide variety of data and languages included in this volume even more. Due to the wide angle of the approach, it was not always clear why certain data or constructions were included.

A broad perspective may include constructions that a linguist working mostly in a generative framework would not include. On the other hand, some questions I would have asked are not addressed. What I found often missing was a discussion of case marking on the P argument, whether it would be treated as oblique or an accusative. Especially considering the common origins of antipassive markers and reflexives, work that looks closer at the exact properties of the case on the P argument would have been appreciated. However, this is probably a concern fueled by personal research interests.

In conclusion, the chapters provide a broad and interesting view of antipassives and how they may be realized in a vast variety of languages. I appreciate this volume and congratulate the editors for providing a frame that, while not limited, gives the reader a frame of reference when such varied constructions are addressed. I always appreciate detailed discussion of primary data from fieldwork, as I am sure readers of Linguistic Typology do too.