CC BY
0
Acta Linguistica Petropolitana. 2024. Vol. 20.2. P. 408-447 DOI: 10.30842/alp23065737202408447
Verb-verb compounds: Delimiting the concept and approaching the study of ordering principles
Alexey I. Vinyar
National Research University Higher School of Economics (Moscow, Russia); [email protected]; ORCID: 0000-0002-8577-9150
Abstract. This study is devoted to verb-verb compounds or monoclausal complex predicates with two or more verbal stems integrated in a single grammatical word. I critically assess earlier approaches to these constructions in terms of their relations to serial verb constructions, a broader family of monoclausal complex predicates. First, I provide a framework to analyze verb-verb compounds and a comparative concept for cross-linguistic study of these structures. Second, based on a small cross-linguistic sample of 12 languages from 6 macro-areas, I present a pilot study of Head-Modifier relations and their ordering principles in three semantic types of these constructions: Manner-of-Motion, Directional and Mental Process Constructions. The findings show that there is no single principle for deriving the component ordering of stems in verb-verb compounds: in Mental Process Verb-Verb Compound Constructions, the ordering follows the general principles of Head-Modifier ordering standard for the morphology and syntax of a given language, while the ordering in Manner-of-Motion and Directional Verb-Verb Compound Constructions the Path Verb is heavily attracted to the position in the end of compound sequence, for which I provide provisional cognitive and diachronic explanations. Different principles governing the Head-Modifier ordering in different types of verb-verb compounds suggest that these compounds do not represent a universal framework. Moreover, a comparison with the principles formulated for Head-Modifier ordering in serial verb constructions in earlier studies suggests a functional affinity of some types of verb-verb compounds and some types of serial verb constructions with less close-knit components.
Keywords: verb-verb compounds, serial verb constructions, heads and modifiers, motion verbs, compounds, incorporation.
Acknowledgments. The article was prepared in the framework of a research grant funded by the Ministry of Science and Higher Education of the Russian Federation (grant ID: 075-15-2022-325). This paper follows my BA thesis on which
© Alexey I. Vinyar, 2024
I was lucky to work under the supervision of Marine Vuillermet. I express my gratitude to Prof. Vuillermet, to Antoine Guillaume, and to Mikhail Alexandrovich Daniel as well to other colleagues in Laboratoire Dynamique du Langage who made my study and stay in Lyon in 2021 both beneficial and pleasant. My special thanks also go to my colleagues from the HSE University: Tatiana Kazakova, Daria Ignatenko, and Ekaterina Vladimirovna Rakhilina, as well as to the anonymous reviewers who made this publication possible.
Глагольно-глагольные компаунды: границы понятия и пилотное исследование порядка компонентов
А. И. Виняр
Национальный исследовательский университет «Высшая школа экономики» (Москва, Россия); [email protected]; ORCID: 0000-0002-8577-9150
Аннотация. Исследование посвящено глагольно-глагольным компаундам, а именно моноклаузальным сложным предикатам, в которых две или более основ глагола объединены в пределах морфосинтаксического слова. В статье приводится обзор предыдущих подходов к схожим конструкциям, а также рассматривается отношение между глагольно-глагольными компаундами с сериальными глагольными конструкциями. На основании предложенного сравнительного понятия я провёл пилотное типологическое исследование принципов, влияющих на порядок компонентов в глагольно-глагольных компаундах. На материале небольшой типологической выборки (12 языков из 6 макроареалов) удалось показать, что порядок компонентов в конструкциях с глаголами перемещения не зависит от линеаризации вершин и зависимых в морфологии и синтаксисе языка, в то время как порядок компонентов в компаундах с глаголами ментальной активности гармонирует с порядком вершины и зависимого в других конструкциях. В конструкциях с глаголами перемещения порядок компонентов в свою очередь определяется тенденцией глагола перемещения занимать последнюю позицию. Принципы, влияющие на порядок компонентов в рассмотренных типах компаундов, схожи с принципами, влияющими на порядок глаголов в менее тесно связанных сериальных глагольных конструкциях.
Ключевые слова: глагольно-глагольные компаунды, сериальные глагольные конструкции, вершины и зависимые, глаголы перемещения, инкорпорация.
Благодарности. Статья подготовлена в рамках гранта, предоставленного Министерством науки и высшего образования Российской Федерации (соглашение о предоставлении гранта № 075-15-2022-325). Эта статья излагает некоторые из результатов моего бакалаврского диплома, который мне посчастливилось писать под руководством Марины Вуйэрмэ во время стажировки в Laboratoire Dynamique du Langage в Лионе. Я благодарен моей научной руководительнице, профессору Вуйэрмэ, а также Антуану Гийому и Михаилу Александровичу Даниэлю за возможность проводить эту научную работу в 2021 году. Кроме того, я хочу поблагодарить Татьяну Казакову, Дарью Игнатенко и Екатерину Владимировну Рахилину, моих коллег из НИУ ВШЭ, помогавших мне на разных этапах подготовки статьи. Наконец, я благодарен анонимным рецензентам и редакторам статьи за их терпение и внимательность.
1. Verb-verb compounds: Comparative concept and a relation to serial verb constructions
In this paper I discuss verb-verb compounds (hereafter, VVCs, see (3), (4) below), a subtype of Serial Verb Constructions (hereafter, SVCs, see (i)-(iii) below), in which two verbal stems form a close-knit unity treated as a single grammatical word. The study has two aims. In terms of methodology, it offers some principles for formulation of comparative concepts needed to study VVCs in light of other types of SVCs and complex predicates. The second aim is to study the ordering principles in the three subtypes of VVCs using the framework distinguishing VVCs on semantic basis. For this purpose, I undertake a pilot study based on a small sample of 12 languages from 6 macro-areas.
The paper is structured as follows. Section 1 is devoted to the problem of distinguishing VVCs from other SVCs; this section also lists the comparative concepts I use and provides a general outline of the pilot study. Section 1.2 specifically addresses the problematic question of sin-gle-wordhood of a construction and its elements' verbhood, while Section 1.3 addresses the notion of headedness in SVCs and VVCs. Section 1.4 presents the language sample used in this pilot study. Section 2
discusses the semantic types of VVCs addressed: Manner-of-Motion (Section 2.1), Directional (Section 2.2), and Mental Process (Section 2.3) constructions. In Section 3, I discuss the component order in the constructions under study and propose some principles which may lie behind their ordering patterns. I summarize my findings and suggest? a further discussion in Section 4.
A serial verb construction, illustrated in (1), (2) below, is a construction with two verbal elements functioning as a single predicate:
Eastern Kayah, Sino-Tibetan family
(1) ?a de the dr plo ku dr ph/s khu
3 put go.up at:u box in at:u shelf on
'They put it (up) in a box on a shelf' [Solnit 1997: 73].
Hoan, Kxa family (Africa)
(2) \ma Wkoe na ka Whoam-Whoam ca
1sg still itin sub jog come
'While I still was coming jogging' [Collins, Gruber 2014: 169].
Serial verb constructions (SVCs) have received much attention in the last forty years. Although SVCs were initially described and analyzed for certain African [Stewart 1963] and East Asian [Li, Thompson 1973] languages with little or no morphology, cross-linguistic studies of SVCs in typologically-diverse languages followed these as well; see [Foley, Olson 1985; Durie 1997; Aikhenvald, Dixon 2006]. These studies acknowledged the existence of SVCs as a family of constructions with multiple variation parameters and subtypes. Among others, the latter feature verbverb compounds, i.e. constructions in which two verbal stems both function as a single clause and form a single verbal word, see [Foley, Olson 1985; Durie 1997]. Consider examples of verb-verb compound constructions in (3)-(4):
Chukchi, Chukotko-Kamchatkan family
(3) t3-kslaw3-pker-Y?a-k ramaj-ets
1sG.s/A-run.slowly-arrive-TH-1sG.s village-DAT
'I came to the village (e.g. from tundra) running slowly' (personal
fieldnotes).
SÄLIBA, ÄUSTRONESIAN FAMILY
(4) ye-tu-dobi-ei-0
3sG-throw-go.down-APP-3sG.o
'He threw it down' [Margetts 1999: 126].
There exist two main definitions of serial verb constructions: an inclusive, prototype-based definition proposed in [Durie 1997: 289-292], summarized in [Aikhenvald, Dixon 2006] and presented in (i) below; and a restrictive comparative concept proposed in [Haspelmath 2016] and cited in (ii). For the comparative concept notion, see [Haspelmath 2010] among others.
(i) "Serial verb construction is a sequence of verbs which act together as a single predicate, without any marker of coordination, subordination or syntactic dependency of any other sort" [Aikhenvald, Dixon 2006: 1].
(ii) "A serial verb construction is a monoclausal construction consisting of multiple independent verbs with no elements linking them and with no predicate-argument relation between the verbs'" [Haspelmath 2016].
Under both approaches, the wordhood of SVC 1 components is viewed as a parameter of variation between languages or between constructions within a single language. Hence, VVCs are treated as a subtype of contiguous SVCs in which two verbal components constitute a single wordform, see [Aikhenvald 2006a]. Consider examples (5), (6) from two closely-related Gunwinyguan languages: Dalabon and Bininj Gun-Wok (Kune dialect). In the Kune dialect of Bininj Gun-Wok, the combination of Manner and Motion verbs is expressed via a contiguous SVC (5), while in Dalabon a VVC is used in the same function (6), see [Evans 2003: 547].
1 Haspelmath [2016] does not view the degree of phonological and morphosyn-tactic boundedness between the components of SVCs as an important typological parameter because of the current lack of universally applicable criterion for wordhood, see [Haspelmath 2011].
Kune dialect of Bininj Kun-Wok (Gunwinyguan, Australia)
(5) kun-dulk nakkanj ka-warme ka-re iv-stick that 3-float.NP 3-go.NP
'A stick is floating along (down the river)' [Evans 2003: 659].
Dalabon (Gunwinyguan, Australia)
(6) dulh djakih kah-warme-ye-bo-n stick that 3-float-ivF-go-PR
'A stick is floating along (down the river)' [Evans 2003: 547].
The functional similarity between VVCs and multi-word SVCs, illustrated for two closely related languages in (5) and (6), was also highlighted for some other languages: e.g., for Yimas (Lower Sepik-Ramu family, Papunesia), see [Foley, Olson 1985], or for Alamblak (Sepik family, Papua area) and Igbo (Niger-Congo, Africa), see [Durie 1997]. Moreover, multi-verb SVCs were demonstrated to be a diachronic source of VVCs for some languages, see [Aikhenvald 1999] for Tariana (Arawak family, South America).
In my study, I also treat VVCs as a subtype of SVCs. However, I propose my own comparative concept for SVCs (v) and VVCs (i) which combines definitions from [Haspelmath 2016] with insights from [Aikhenvald, Dixon 2006]:
(iii) my definition of the SVC:
A serial verb construction is a monoclausal construction consisting of multiple verbs with no marker of coordination or subordination between them.
(iv) my definition of the VVC:
A verb-verb compound construction is a serial verb construction whose elements can be separated only by non-word-class-changing derivational elements.
In the following subsections, I provide a more detailed discussion of the features differentiating VVCs from other SVCs (Section 1.1) and of the types of VVC components (Heads and Modifiers, Section 1.2).
1.1. Serial verb constructions and verb-verb compounds: A problem of verbhood and single-wordhood
To study verbal compounds separately from other types of SVCs, I need a universally applicable criterion for wordhood. Unfortunately, the existing studies on compounding assume that there is no universal way to differentiate between compounds and phrases: the phonological, morphosyntactic and semantic criteria proposed are often language-specific and do not converge even within a single language (e.g., see [Lieber, Stekauer 2009]. Moreover, typol-ogists currently lack a universally applicable comparative concept for the notion of word [Haspelmath 2011]. Because a cross-linguistic analysis of word-hood is beyond the aims of my research, the proposed VVC definition is rather restrictive and based on a single criterion of contiguity and separability.
Considering that my study focuses on VVCs, I can make Martin Haspel-math's definition of SVC [Haspelmath 2016] less restrictive: Haspelmath introduces "no linking element" and "no predicate-argument" relation to exclude multi-verb constructions that can be potentially analyzed either as bi-clausal or as SVCs if no additional information is provided (7). However, constructions like (7) can be excluded from SVCs by the criterion of monoclausality.
(7) She helped me solve the problem [Haspelmath 2016].
This restrictiveness is redundant for my purposes: members of contiguous compounds (VVCs) cannot in general be analyzed as belonging to different clauses. 2 Making this comparative concept more inclusive has its benefits: in some languages, members of VVCs have to be linked by a 'dummy' morpheme because of morphophonological considerations.
To specify my definition of a 'dummy' morpheme and a 'derivational' element, a given morpheme is considered here as dummy if it only occurs between the components of compounds and hence cannot be treated as a marker of coordination or subordination. According to this approach,
2 Such an analysis, however, is by no way impossible (see the discussion of various cases in [Panova 2017 ms]). However, the maj ority of such spurious cases can be analyzed by direct application of Haspelmath's criterion of monoclausality [Haspelmath 2016].
I classify linking morphemes in Dalabon (see the morphemes glossed as 'ivf' (incorporated verb form) in (6) above) and Bininj Kun-Wok as dummy rather than as subordination/coordination markers. At the same time, I do not regard sequential multi-verb constructions in Yimas (Lower Sepik-Ramu family, Papunesia) as SVCs (and, consequently, as VVCs). According to Foley (1991: 325-326), verbal components of these constructions can be separated by the suffix -mpi (8) which also marks dependent verbs.
(8) arm-n kay i-ka-ak-mpi-wul
water-oBL canoe.vm.sG vm.sG.o-1sG.A-push-sEQ-put.down
'I pushed the canoe down in the water' [Foley 1991: 326].
By derivational elements, I understand non-inflectional bound morphemes. To identify inflectional morphemes, I follow [Bickel, Nichols 2007] and [Bickel, Zuniga 2017]. Informally speaking, an inflectional morpheme is a morpheme whose presence is required by a stem of a particular class to be used in a particular context.
Consider two types of SVCs in Tlachichilco Tepehua (Totonak family, North America) — Level I and Level II SVCs as they are named in [Watters 1988]. The constructions differ in two ways, see [Watters 1988: 8791]. First, in Level II SVCs, an epenthetic -h- is inserted between the components of a VVC if the first component ends in a vowel or /n/ consonant
(9). In Level I SVCs, an epenthetic -h- is inserted only if the first element ends in an Antipassive suffix; compare (10) and (11). Second, in Level
II SVCs, an inflectional prefix can occur between the SVC components, see (12) [Watters 1988: 91], while in Level I SVCs, an inflectional prefix can only precede both components, see (13) [Watters 1988: 90-91].
Level II SVC, Tlachichlco Tepehua
(9) milpa-h-wi:l
sing-EPHEN-sit
'Sit singing' [Watters 1988: 88].
Level I SVC, Tlachichilco Tepehua
(10) st'a-la:'an-ta sell-take-PF
'X goes selling Y' [Watters 1988: 88].
Level I SVC, Tlachichlco Tepehua
(11) 'oq-nuh-min-ta
drink-AP-come-PF
'X is coming, drinking' [Watters 1988: 95].
Level II SVC, Tlachichilco Tepehua
(12) milpa:-h-lak-tawla-ni-y
sing-EPHEN-3PL.OBJ-sit.down-DAT-IMPFV
'X sits by them (his younger siblings) singing' [Watters 1988: 91].
Level I SVC, Tlachichilco Tepehua
(13) a. lak-st'a:-li:.mm-ta pu:laqli
3pL.oBj-sell-come-PF tamales
'X comes selling tamales' [Watters 1988: 90].
b. *st'a:-lak-li:.mm-ta pu:laqli
sell-3pL.oBj-come-PF tamales
Expected: 'X comes selling tamales' [Watters 1988: 90].
In consistence with my definition, I do not treat Level II SVCs in Tlachichilco Tepehua as VVCs, because they can be separated by an inflectional morpheme. I consider Level I SVCs as an instance of VVCs in this language, because the components of these VVCs can only be separated by means of derivational morphology; cf. (11) and (13).
Finally, a note about monoclausality is needed here. Following [Bohn-meyer et al. 2007] and [Haspelmath 2016], I regard the absence of independent negation of construction components as a criterion for mono-clausality. In many cases, this criterion is concordant with my criterion of non-separability, whereby components of strictly contiguous VVCs cannot be negated independently. However, some languages can use derivation as a means of expressing negation. For example, in Kwaza (unclassified, South America), the negative morpheme -he can occur without any host [van der Voort 2004: 527-530] and hence can be regarded as derivational. While multi-verb constructions in Kwaza appear to instantiate a single grammatical and phonological word [van der Voort 2004: 567568], the negation can be applied to each component of a multi-verb construction independently (14), see [van der Voort 2004: 531-532].
Kwaza, unclassified (South America)
(14) ja-'he-kui-'he-tse
eat-NEG-drink-NEG-DEc
'He didn't eat and didn't drink' [van der Voort 2004: 532].
To be defined as an SVC or VVC, a given construction should consist of verbs. For [Aikhenvald, Dixon 2006], it is sufficient that a component of a given construction can appear in the same form functioning as a predicate of an independent clause. However, as noted by [Enfield 2009], the verbal status of some SVC components can be questioned: even if a SVC component can appear in the same form as a single predicate, its semantic contribution to the meaning of the SVC can differ drastically from its meaning outside the SVC. The same puzzle holds for VVCs as well. Consider examples (15), (16) from Mapudugun (Ar-aucanian, South America).
(15) kupa-ulkantu-n
come-sing-1sG.iND
'I want to sing' [Zuniga 2017: 706].
(16) kupa-nu-a-fu-lu rangi antu ama
come-NEG-NRLD-iPD-svN mid day part
'Wasn't he supposed to come at noon' [(Smeets 2007: 338].
In (16), kupa 'to come' is used as an independent verb describing the event of coming. However, in (15) it occurs in the VVC as V1 with a drastically different, grammaticalized desiderative meaning. This case shows that the independent occurrence criterion is not sufficient: it is probably better to regard kupa not as a V1 in the VVC, but as a grammaticalized derivational desiderative morpheme (homophonous to the kupa verbal root), attached to another verbal root 3. Otherwise, many cases of construction-specific grammaticalization of verbal stems in which kupa has
3 This consideration is purely typologically-comparative: the suffixing grammar of Mapudungun language makes it difficult to postulate a verbal derivational prefix here. However, it is hardly desirable to incorporate such language-specific criteria into typological research.
lost any transparent semantic relation with the verbal lexeme would have to be defined as a type of VVC. Though the latter approach is not entirely scientifically impossible, it might be more relevant for a study addressing non-compositional VVCs as well.
Haspelmath [2016] excludes grammaticalized SVCs such as (15) from his SVC concept by introducing the 'independent verb' and 'construction' notions in his comparative concept (ii). According to Haspelmath, the independent verb must be able to '... express a dynamic event without any special coding in predication function (...) and can occur in a non-elliptical utterance without another verb'. For a given construction to be considered as SVC, Haspelmath requires that '.(it) must be a productive schematic (construction)' and that '... the meaning of a concrete construct can be determined based on the meanings of its parts and the construction meaning...'.
The approach advocated in [Haspelmath 2016] makes it possible to exclude cases such as (15) from the list of SVCs. However, it will also exclude the majority of asymmetrical constructions in the sense of [Aikhenvald 2006]. Moreover, as noted in [Enfield 2009], it is often difficult to distinguish between the construction meaning and the meaning of each of its components. Consider Directional VVCs (17)-(20) from Mapudu-ngun [Zuniga 2017: 706-707]:
(17) anu-pura-i
sit-ascend-iND 'He sat up'.
(18) anu-nag-i
sit-descend-iND
'He sat down'.
(19) rungku-tripa-i
jump-exit-iND
'He jumped out'.
(20) rungku-kon-i
jump-enter-iND
'He jumped in'.
In these VVCs, the Motion verb which occupies the V2 slot indicates the Direction of the action expressed by the verb in the V1 slot. Now consider these Motion verbs used as independent verbs in (21)-(24):
(21) pura-pura-kunu-w-nge!
go.up-go.up-SFR-PRPS-REF-IMP2s
'Get upstairs quickly' 4 [Smeets 2008: 307].
(22) epe kon-un antu. almost enter-pvN sun
'When the sun had almost set...' [Smeets 2008: 403].
(23) tripa-ke-y-ng-un pun... leave-cF-iND-3NS-p night
'They go out at night.' [Smeets 2008: 352].
(24) ella naq-un antu... a.bit go.down-pvN sun
'When the sun went down a bit.' [Smeets 2008].
Do these verbs in (17)-(20) mean the same as in (21)-(24)? To what extent is the event of Motion present when these Motion verbs are used as V2s in VVCs? If we state that the Motion verbs lose the Motion component of their semantics in Directional VVCs, then we should not consider (17)-(20) as VVCs. However, Directional VVCs in Mapudungun differ from the case of the grammaticalized verb kupa 'come' used as a Desiderative marker.
To solve the problem of possible mismatches between the meaning of a verb in its independent and constructional usage, I suggest the following concept of a verb in SVC/VVC constructions, partly based on [Haspelmath 2016]:
(v) To count as a verb, the verbal element must make at least some lexical contribution to the semantics of the whole construction. Moreover, a verbal element in an SVC/VVC can be considered a lexical verb only if it contributes either more or less semantic content
4 Reduplication indicates that the action is performed quickly [Smeets 2008: 307].
(compared to its meaning as a single predicate) to the meaning of the whole construction, though not more AND less simultaneously.
While this definition excludes fully grammaticalized VVCs like the one in (15), it still treats as VVCs cases of construction-dependent behavior of verbs in certain VVCs like those in (17)-(20).
Finally, I would like to discuss the formal mismatch between verbs used as VVC components and those used as independent verbs. Apart from the general phonological rules of a given language, specific morph-ophonological rules can apply to VVCs. Where the mismatches emerging due to such rules are at least partially predictable, in my study I still consider the verbal elements that can occur both as VVC components and as independent predicate roots to represent the same verbal entities. This is the case of the 'incorporated verb forms' in Dalabon, see (6) and the corresponding reference. In some languages, however, independent verbs and VVC components sharing the same meaning can drastically differ phonologically. Consider (25) from Tiwi (isolate, Australia):
(25) ji-msni-marikuayspi-ya
HE-ME-dancing-grab
'He grabbed me while I was dancing' [Osborne 1974: 47].
The form marikuaqspi describing the dancing event is only found in VVCs, while its free-verb synonym is joi 'to dance' [Osborne 1974: 4748]. All verb-headed compounds in Tiwi exhibit such strong and unpredictable suppletive patterns (see the list of forms in [Osborne 1974: 4849]). In my study, I do not treat such compounds as VVCs. The same holds for "affixal predicates" which can be hosted by verbal stems in Wakashan languages: see [Wojdak 2008: 159-169] for Nuu-chah-nulth.
1.2. Heads and Modifiers in SVCs/VVCs
With regard to Head-Modifier relations in SVCs, Aikhenvald [2006] establishes two broad classes of SVCs defined by the compositional and semantic properties of their components [Aikhenvald 2006: 22-23].
Verbal components in symmetrical SVCs come from an open class, have the same impact on the semantics and morphosyntax of the whole construction and tend to lexicalize [Aikhenvald 2006: 28-36]. By contrast, components of asymmetrical SVCs differ in this respect: "minor" verbs come from a closed class and can only modify the "major" (open class) verb meanings. "Minor" verbs tend to be semantically bleached and are prone to grammaticalization [Aikhenvald 2006: 30-36]. All asymmetrical SVCs are semantically headed by the major verb, while only some symmetrical SVCs are semantically headed. The mutual Head — Modifier ordering in semantically headed constructions is construction-specific.
VVCs with verbs hipah 'know how to' and tuk 'want to' in Hup (Na-dahup family, South America) can be viewed as typical instances of asymmetrical VVCs, see [Epps 2008: 420-421]. Hereafter, I define such structures as 'Mental Process VVCs'. These and several other event-argument-taking verbs (viewed on the semantic, but not on the syntactic plain) can directly follow their verbal arguments in VVCs as in (26), (27). Hence, we can conclude that the V2 slot in the Mental Process VVC in Hup is 'closed', while the V1 slot is 'open'.
Hup, Nadahup family
(26) yit=mah tih [yo-d'o?]-hipah-nih
thus-REp 3sg [hang.from.above-take]-know-NEG
g'5h-g'et-g'o?-op=b'ay
be-stand-go.about-DEp-AGAiN
'So he was standing around, not knowing how to carry (the fish)'
[Epps 2008: 420].
Hup, Nadahup family
(27) ?in-an [bi-hitam]-tuk-yo?...
1pl-obj [work-cooperate]-want-SEQ
'Having wanted to help us' [Epps 2008: 420].
Hup shows an example of a typical symmetrical VVC as well. According to Epps [2008: 415-416], an unrestricted set of verbs can directly precede another verb to express an action performed simultaneously with
that of the second verb as in (28), (29). This construction exhibits no evidence of semantic bleaching of any of its members.
(28) ?ah ?3g-g'o?-6y
1sg drink-go.about-DYNM
'I would go around drinking' [Epps 2008: 411].
(29) tih yamhido?-g'5p-oh
3sg sing-serve-DECL
'I was singing while serving beer' [Epps 2008: 416].
Aikhenvald's [2006] approach is prototype-based, and she is aware that her proposed classification is not universally applicable [Aikhenvald 2006: 36]. As noted by [Enfield 2009], a class can be closed or open only compared to another class. Consider the VVC in Ese Ejja (Takanan family, South America) in (30), (31). In this VVC, the first verb expresses the Caused Motion event, while the second specifies its Direction; see Section 2 for definitions of the constructions addressed.
(30) ye-kwaya-ka-a=kwana=pwa bring-put.out-3A-RPAS=3PL=RPAS
'We went out (by river Natawa) (lit. The river brought us.)' [Vuillermet 2013: 420].
(31) inawewa=pi'ai kwiji-kwiji-ani
dog-ALSO bark-RED-PRS
ojaya owa=zapato jya-sowa-ka=jo 3gen 3ERG-shoe(Sp) throw-go.up.TR-3A=TMP.DS
'The dog barks when he lifts his shoe' [Vuillermet 2013: 579].
The V1 slot is restricted to Caused Motion verbs and the V2 slot is restricted to several Path verbs. Both slots in this construction should be considered as closed according to [Aikhenvald 2006]. However, it is unclear whether the Directional VVC in Ese Ejja and the Mental Process VVC in Hup are asymmetrical in the same sense.
Moreover, the symmetrical/asymmetrical distinction is based on heterogeneous parameters (semantic and morpho-syntactic headed-ness and compositional restrictions on each component). If we consider
again the Mental Process VVC in Hup, we can conclude that the semantic head is located in the V2 slot. However, the list of verbs which can enter the V2 slot is more restricted than the list of verbs that can enter the V1 slot.
Despite the shortcomings discussed above, it is tempting to preserve and test the generalizations established in the framework of [Aikhen-vald 2006]. To make symmetrical/asymmetrical distinction more universally applicable and fine-grained, I propose to study semantic head-edness independently of compositional properties of elements. As noted in [Croft 2001], the relationship between Heads and Modifiers can only be studied cross-linguistically by employing a universally applicable semantics-based definition of a Head. Such a definition was provided by [Zwicky 1985: 4] and elaborated in [Croft 2001: 257] in his concept of Profile equivalent (vi):
(vi) In a combination X + Y, X is the profile equivalent if X profiles/describes a kind of thing profiled/described by X + Y.
I use Croft's definition of profile equivalent as a definition of Head in VVCs. At the same time, while, according to this definition, Head cannot be established in every construction, in my analysis, I consider only those VVCs that have an identifiable semantic Head (see Section 2).
1.4. A pilot sample
I use a sample of 12 languages (Table 1, p. 424), the language and family names taken from Glottolog, except for =IHoan which is indicated as Amkoe in Glottolog.
Though small, this sample is genetically and areally balanced, see [Hammarstrom, Donohue 2014]. The only aspect where it can be considered unbalanced is that this study is focused on three VVCs: the Man-ner-of-Motion VVC, the Directional VVC, and the Mental Process VVC (see Section 2), with the latter VVC being not only the least frequent of them but also absent from languages from 3 macroareas: Multinesia, Australia, and Africa (see Table 2, p. 424).
Table 1. Language sample
Language Family Macroarea
Hup Nadahup South America
Mapudungun Araucanian South America
Ese Ejja Takanan South America
Tlachichilco Tepehua Totonac North America
Kiowa Kiowa-Tanoan North America
Chimalapa Zoque Mixe-Zoque North America
Eastern Kayah Sino-Tibetan Eurasia
Chukchi Chukotko-Kamchatkan Eurasia
Saliba Austronesian Multinesia
Imonda Border Multinesia
Bininj Kun-Wok Gunwinyguan Australia
=IHoan Kxa Africa
Table 2. VVCs and macroareas
Language Manner-of-Motion Directional Mental Process Macroarea
Hup + + + South America
Mapudungun - + + South America
Ese Ejja - + - South America
Tlachichilco Tepehua + - - North America
Kiowa - - + North America
Chimalapa Zoque - + + North America
Eastern Kayah + + + Eurasia
Chukchi + - - Eurasia
Saliba + + - Multinesia
Imonda + + - Multinesia
Bininj Kun-Wok + + - Australia
=IHoan + + - Africa
The data for each language is taken from reference grammars and, where possible, from publications on this specific topic. The data for the Chukchi language was partially acquired during my fieldwork in 2016-2017. For Tlachichilco Tepehua and Bininj Kun-Wok, data from sister languages was additionally used for comparative purposes (these sources are also indicated in Table 3). The data sources I used are represented in Table 3.
Table 3. Reference data
Language Sources
Hup Epps 2008
Mapudungun Zûniga 2006; 2017; Baker et al. 2005; Smeets 2008
Ese Ejja Vuillermet 2012; Vuillermet 2017
Tlachichilco Tepehua Watters 1988; Kung 2007
Kiowa Watkins 1984
Chimalapa Zoque Johnson 2000
Eastern Kayah Solnit 1997; 2006
Chukchi Dunn 1999; Muravyova et al. 2000
Saliba Margetts 1999
Imonda Seiler 1985
Bininj Kun-Wok McKay 1975; Evans 2003; Evans 2017
=IHoan Collins, Gruber 2014; Berthold, Gerlach 2017
2. Constructions studied
2.1. The Manner-of-Motion VVC
(vii) The Manner-of-Motion VVC
The Manner-of-Motion VVC is a VVC in which one component indicates a Motion event, while another (the other?) component describes the Manner the Motion is performed or, optionally, an Action performed simultaneously with the Motion event.
I follow Talmy [2000] in his definitions of Manner and Motion. In my study, I do not differentiate between Motion verbs that additionally lex-icalize Path and those that do not. Hereafter, I do not use the label "Motion verb" for Manner-of-Motion verbs.
The last component of my definition (vii) needs a special note. Consider the following VVCs from Bininj Kun-Wok (Gunwinyguan, Australia) in (32), (33):
(32) nahnane 0-wage-yihmi-re-i
ma:dem 3p-crawl-ivF-go-pi
(pointing to the tracks left by a Dreamtime being) 'This is where he came crawling along' [Evans 2003: 543].
(33) ga-ganj-ngu-nihmi-re 3-meat-eat-ivF-go.NP
'He goes along eating meat' [Evans 2003: 536].
The last verb in both VVCs is a Motion verb. Notably, however, the semantic class of the verbs occupying the V1 slot is less evident. In (32) it is a Manner verb, and the semantic Head of this construction is clearly a Motion verb. The situation is different in (33). Normally, one does not classify verbs like 'eat meat' 5 with those of the Manner of Motion. How-
5 The V1 in this VCC is itself a complex verb stem with an incorporated nominal stem (the morphological structure and noun incorporation process in Bininj Kun-Wok is described in detail in [Evans 2003]).
ever, in several languages of my sample (Hup, Tlachichilco Tepehua, Sal-iba, Bininj Kun-Wok and =IHoan), these evidently non-Manner verbs describing simultaneous co-events can occur in the same slot with Manner verbs. This makes the distinction between Manner and non-Manner verbs in such languages less clear.
Considering that for many languages the distinction between "Manner co-event" and "Simultaneous Action co-event" is obscure, I regard all the VVCs presented in (32)-(33) as instances of Manner-of-Mo-tion VVCs. For my study, the ability of "genuine" Manner verbs (e.g., 'run', 'jump', 'swim', etc.) to participate in VVCs is crucial. Hence, if a language only shows VVCs structured like 'eat-go' (go while eating), 'sing-come' (come singing) but no VVCs like 'run-go' (go running), 'swim-come' (come swimming), I do not treat VVCs in this language as Manner-of-Motion VVCs.
The semantic Head status of the Motion verb in these constructions is not always evident. Some grammars note that a Motion verb is always(??) a semantic Head (e.g., see [Epps 2008)] for Hup) and reflect this correspondingly in translations (e.g., (32), (33) from Bininj Kun-Wok above). In other cases, the Head-Modifier structure of these VVCs is less clear. Consider (34), (35) from Eastern Kayah (Sino-Tibetan, Eurasia):
(34) pipe jo cwa r<i butterfly fly go 0
'They (butterflies) flew away' [Solnit 1997: 77].
(35) je cwa r<i sine carry.on.shoulder go part gun
'(They) went carrying guns on their shoulders' [Solnit 1997: 73].
The question arises if there are any semantic Heads in these constructions or, in other words, if (34) primarily describes an event of 'flying' or an event of 'going'? According to the English translation in [Solnit 1997: 73], at least the verb 'go' in (35) is the semantic Head. At the same time, Solnit [2006: 150] adds an alternative translation to this example: '(they) carried away guns on their shoulders'. This translation indicates that the verb je 'carry.on.shoulder' is a possible Head and the whole
construction may instantiate a Directional VVC but not a Manner-of-Mo-tion VVC. I address this problem in the following section.
Among the languages included in my sample, Manner-of-Motion VVCs are present in Hup, Tlachichilco Tepehua, Eastern Kayah, Chukchi, Saliba, Imonda, Bininj Kun-Wok and =IHoan.
2.2. The Directional VVC
(viii) The Directional VVC
The Directional VVC is a VVC in which one component indicates an Event and the other component indicates the Direction in which this Event evolves.
The term Direction I use here roughly corresponds to Talmy's [2000] notion of Path. However, unlike Path, Direction can be specified even for an event with no Motion component. Consider Directional VVCs in Saliba (Austronesian, Papunesia):
(36) ye-tu-dobi-ei-0
3sG-throw-go.down-APP-3sG.o
'He threw it down' [Margetts 1999: 126].
(37) ye-hedede-dobi i-wane "Eey Tau Mekemekeya... " 3sG-tell-go.down 3sG-say intrj man Name
'He called/spoke down and said "Eey, Tau Mekemekeya"' [Margetts 1999: 122].
In (36), the Motion verb dobi 'go down' specifies the Direction of a Caused Motion event expressed by the verb tu 'throw'. In (37), however, there is no Motion event—the verb dobi only indicates the downward orientation of the participant.
For ambiguous cases, I classify constructions as either a Man-ner-of-Motion VVC or a Directional VVC depending on the translation provided by the grammar's author. Some cases, however, are not ambiguous: I did not classify as Directional those VVCs where only Manner verbs can combine with Motion verbs. Instead, such VVCs are classified
here as Manner-of-Motion VVCs. Constructions with no motion component cannot be classified as Manner-of-Motion VVCs either. Consider (38)-(40) from Imonda (Border family, Multinesia):
(38) tetoad paiha-i-pia-n bird fly-LiNK 6-come-psT
'The bird came flying' [Seiler 1985: 108].
(39) lol-peha-na-f
talk-go.down-BEN-PRs
'Go down, talking to him' [Seiler 1985: 109].
(40) lol-peha fe-na-f
talk-go.down do-BEN-PRs
'Talk down to someone' [Seiler 1985: 109].
Examples (38) and (39) illustrate a Manne-of-Motion VVC and example (40), a Directional VVC. The semantic difference between (39) and (40) is a key to distinguish between the two VVCs in Imonda.
Among the languages in my sample, Directional VVCs are found in Hup, Mapudungun, Chimalapa Zoque, Eastern Kayah, Saliba, Imonda, Bininj Kun-Wok, and =IHoan.
To conclude this discussion, the distinction between Heads and Modifiers in Directional and Manner-of-Motion VVCs turns out to be virtually irrelevant for their formal properties. As discussed below in Section 3.2, the order of components in these VVCs seems to be influenced by features orthogonal to the Head-Modifier distinction.
6 According to Seiler [1985: 108; 119-132], the only function the linker -i- has in the modern Imonda is that of a morphological link between members of certain VVCs and between the classifier prefix and the verbal root. Notably, classifiers show a historical link with some verbs occurring in VVCs; see [Seiler 1985: 119-130].
2.3. The Mental Process VVC
(ix) The Mental Process VVC
A Mental Process VVC is a VVC in which mental process verbs combine with verbs denoting arguments of the events described.
The only 'mental process verbs' considered in this study are: 'to want' (41), 'to know/ understand' (42) and 'to learn' (43). All the three can be found in VVCs in Chimalapa Zoque (Mixe-Zoquean, North America):
(41) piceqho? dss ds=min-t3?-ke?t-pa thus 1prn 1A=come-want-REPET-iNc
'That's why I want to come back again' [Johnson 2000: 237].
(42) ney komo ?une=dskka kwandu ?otoy-?ay.may-suk-ws same like child=NPL when speak-learn-3pL-coM
'Just like children when they learn to speak' 7 [Johnson 2000: 237].
(43) pwes ?aber hunaq bi kopak si yos-mus-pa well let's.see how def head if work-know-iNc
'Well, let's see how he thinks, if he knows how to work' [Johnson 2000: 311].
Among the languages in my sample, Mental Process VVCs are present in Hup, Mapudungun, Kiowa, Chimalapa Zoque and Eastern Kayah.
3. Component ordering and its governing principles
Alexandra Aikhenvald [Aikhenvald 2006] assumes that the order of components in semantically-headed SVCs is language- and
7 There is probably a misprint in the translation line in [Johnson 2000: 237]: this sentence is translated as 'Just like children when they learn to walk'. However, no points about (non)compositionality are made. Throughout the grammar, V1 ?otoy is invariably translated as 'speak'.
construction-specific. This conclusion is at odds with [Foley, Olson 1986] who note that Motion and Posture verbs tend to occur as final members of SVCs. Moreover, it was argued by Givon [2009] that SVCs (and, consequently, VVCs) arise from multiclausal structures (e.g., see [Rose 2009] for a study of this process in Emerillon). Hence, it can be expected that the Head and Modifier ordering in VVCs would correlate with that in other constructions. Finally, even if the order of components in a given construction is language-specific, it does not necessarily mean that this order is random cross-linguistically. In my pilot study, I investigate potential correlations between the order of Heads and Modifiers in VVCs and the features listed below in (viii). The correlations are addressed independently for 3 types of VVCs.
(ix) Features that possibly correlate with the component order in VVCs:
— The dominant verb — object order.
— The order of clauses in semantically correlative analytical constructions (if any).
— The Head — Modifier order in other verb-headed compounds (hereafter, incorporation) as, e.g., in noun incorporation constructions (where other compound constructions exist).
I use Matthew Dryer's [Dryer 2013] principles of determining the dominant object—verb order in a language. Where the order is flexible, this is indicated in the corresponding section of Table 4 below.
Recall that I take into account here not only the object and verb ordering but also the ordering in multi-verb syntactic structures that are (quasi)synonymous to the VVCs addressed. It should be noted that only Manner-of-Motion and Mental Process VVCs have semantically obvious corresponding syntactic structures while multi-verb syntactic structures that express (roughly) the same meaning as Directional VVCs are difficult to find.
Manner-of-Motion VVCs can be (quasi)synonymous with a wider range of syntactic structures. The closest equivalent for Manner-of-Mo-tion VVCs in Bininj Kun-Wok (Gunwinyguan) is a multi-word SVC. At the same time, Manner-of-Motion VVCs correspond to combinations of finite verb and converb clauses in Hup (Hupde) and Chukchi
(Chukotko-Kamchatkan). To avoid methodological opportunism, I did not compare the order of components in Manner-of-Motion VVCs with that in structures (quasi)synonymous with these VVCs.
This paper takes the biclausal structure of [MATRIxMental Process verb [EMBEDDEDComplement verb]] to be corresponding to that of Mental Process VVCs. Compare a Mental Process VVC in (44) and, presumably, a (quasi) synonymous syntactic structure in (45) from Chimalapa Zoque (Mixe-Zo-quean, North America):
(44) piceqho? dss ds=min-ts?-ke?t-pa Thus 1prn 1A=come-want-REPET-iNc
'That's why I want to come back again' [Johnson 2000: 237].
(45) dss ?sn=ts?-pa yak-csn-ws kastro lo cruz 1prn 1E=want-iNc cAus-sit-coM castro lo cruz
'I want Castro lo Cruz to be elected' [Johnson 2000: 308].
In this study, I compare the order of components in VVCs both to that of Object — Verb, Head — Complement Verbs, and to the ordering of components in noun incorporation (NI) constructions, see [Mithun 1984].
In general, NI is considered to be an optional construction, see [Mithun 1986]. Building on its definition in [Caballero et al. 2008], I propose another definition in (x) below. Note that my definition applies to languages that lack verbal inflection and excludes cases of 'obligatory incorporation'.
(x) A definition of NI constructions.
NI is an optional construction used to express a particular semantic relation between the predicate and one of its participants. A NI construction combines the verb (which expresses the predicate) and the nominal element (which expresses a participant) in a single NI construct. The incorporated nominal element (hereafter, IN) cannot perform more discourse functions than those it performs in alternative constructions. Only derivational elements can separate the IN from the verb. No inflectional elements pertaining to the nominal component or its modifiers can occur inside the verbal construct.
To specify my definition of an optional construction, optional means here that a given NI construction is not the only one a given language can use to express a given relation between the Verb and its Patient, the Verb and its Instrument, etc.
Of course, this definition has its limitations. In consistence with (x), Lakhota (Siouan, North America) and some Athapaskan languages like Koyukon (Na-Dene, North America) should lack NI constructions altogether. However, these and other methodological controversions fall outside the scope of my VVC pilot study.
Presented below is data on the Head-Modifier ordering in the three VVCs under study and in comparable constructions found in all the languages of my sample (Table 4). The subsequent subsections discuss the findings of the comparison and provide some generalizations with likely explanations.
Table 4. Ordering in core sample
Language Manner-of-Motion Directional Mental Process OV/VO Incorporation
Hup M-H M-H M-H OV N-V
Mapudungun H-M H-M VO V-N
Ese Ejja H-M OV N-V
Tlachichilco Tepehua M-H VO N-V
Kiowa M-H OV N-V
Chimalapa Zoque H-M M-H VO N-V
Eastern Kayah M-H H-M M-H VO V-N
Chukchi M-H OV N-V
Saliba M-H H-M OV V-N; N-V
Imonda M-H H-M
Bininj Kun-Wok M-H M-H OV N-V
=IHoan M-H H-M VO
3.1. Mental Process VVCs
Mental Process VVCs are found in 5 languages of my sample, with 2 OV and 3 VO languages among them. My observation shows the Mental Process VVC to be the only VVC exhibiting strong correlation between the order of Head and Modifier components and the order of constituents in other constructions (Table 5).
Table 5. Mental Process VVCs ordering
H-M order OV / VO Inc Complement
Hup M-H OV NV C-H?
Kiowa M-H OV NV no data
Mapudungun H-M VO VN H-C
Chimalapa Zoque M-H VO NV; AdvV C-H?
Eastern Kayah H-M VO VN C-H
Chimalapa Zoque is the only language in which the order of Heads and Modifiers in this VVC does not agree with the order of verb-and-object and head-and-complement clauses. At the same time, the Head-and-Modi-fier ordering in Chimalapa Zoque Mental Process VVCs is harmonic with that of Heads and Modifiers in incorporation constructions. While it is difficult to make any generalizations on the basis of a single language, I argue that Chimalapa Zoque represents a telling case. According to Campbell, Kaufman, Smith-Stark [1986: 54], Mixe-Zoquean languages are likely to have displayed the OV order before their shift to the VO order under the influence of the Mesoamerican linguistic area. Hence it is possible to regard the Modifer-Head order in Chimalapa Zoque Mental Process VVCs (together with the N-V and Adv-V order in incorporation constructions) as a relic of its historical OV order.
I argue that the correspondence between the order of Heads and Modifiers in Mental Process VVCs and the order of Heads and Modifiers in incorporation constructions is of a diachronic nature. The order of components in the Mental Process VVC reflects the order
of head-and- complement clauses in these languages that existed at the time this VVC arose; (see similar considerations for nominal compounds in [Comrie 1980: 85] and [Gaeta 2008: 122].
3.2. Manner-of-Motion and Directional VVCs
Manner-of-Motion VVCs are found in 9 languages of my sample, with 6 of these exhibiting the dominant OV order and 3, the VO order. Manner verbs precede Motion verbs in all the languages addressed (Table 6). Hence, the order of Head and Modifier in Manner-of-Motion VVCs does not agree with the object and verb ordering in all three VO languages. If, however, the ordering in incorporation constructions is also taken into account, this disharmony is only true for Eastern Kayah (|=Hoan lacks incorporation). Recall that Saliba displays both N-V and V-N ordering in incorporation construction.
Table 6. Ordering in Manner-of-Motion VVCs
VVC OV / VO Inc
Hup M-H OV N-V
Ese Ejja M-H OV N-V
Chukchi M-H OV N-V; Adv-V
Saliba M-H OV N-V; V-N
Imonda M-H OV —
Bininj Kun-Wok M-H OV N-V; Adv-V
Tlachichilco Tepehua M-H VO Adv-V
Eastern Kayah M-H VO V-N
|=Hoan M-H VO —
Directional VVCs are found in 9 languages of my sample. Except for two OV languages (Hup and Bininj Kun-Wok), this VVC exhibits the Head-Modifier ordering. Hence, 3 OV languages demonstrate disharmony between the VVC and the in-clause ordering (as shown in bold in Table 7).
Table 7. Directional VVCs
H-M order OV / VO Inc
Hup M-H OV NV
Bininj Kun-Wok M-H OV NV; AdvV
Ese Ejja H-M OV NV
Saliba H-M OV NV; VN
Imonda H-M OV —
Mapudungun H-M VO VN
Chimalapa Zoque H-M VO NV; AdvV
Eastern Kayah H-M VO VN
|=Hoan H-M VO —
Unlike in Mental Process VVCs, the order of components in Directional and Manner-of-Motion VVCs does not seem to strongly correlate with the ordering of Heads and Modifiers in other constructions addressed. Discussed below are some speculative considerations, to be elaborated in forthcoming studies, on the factors restricting the component ordering in these VVCs.
Recall that the Head and Modifier status of components in Man-ner-of-Motion VVCs is problematic. Moreover, the difference between Man-ner-of-Motion VVCs and Directional VVCs is not obvious for languages where both constructions are present. Hence, it is probably safer to discuss the ordering of Path-encoding verbs (which function as Directional verbs 8 in Directional VVCs and as Motion verbs in Manner-of-Motion VVCs) vs. non-Path-encoding verbs in VVCs. Non-Path-encoding verbs in Man-ner-of-Motion VVCs function as either Manner-of-Motion verbs or verbs linking simultaneous co-events to Motion events. Path verbs tend to follow non-Path verbs in Manner-of-Motion and/or Directional VVCs.
SVCs are frequently discussed with respect to the iconicity of their components' ordering and the sequential relation between the (sub)events
8 Recall that Motion verbs in Directional VVCs are not necessarily Path verbs in Talmy's [2000] sense; according to my definition, they can also function as Deictic verbs.
they express; see [Durie 1997; Bril 2004; Aikhenvald 2006]. One of the issues addressed is whether it is possible to directly attribute the [non-Path verb]-[Path verb] preferred order to any cognitive factor. My answer is 'no': no cognitive explanation can be directly applied to the order preference observed in my sample.
The most obvious argument against cognitive motivation of the [Man-ner]-[Motion] and [Action]-[Direction] ordering is that the latter seems to be preferred in only SVCs and VVCs. As regards bound Direction/Path morphemes, one can find languages with prefixal Path expression and Manner following Motion in verb-framed languages; see the languages addressed(??) in [Slobin 2004]. I am unware of any studies claiming the existence of any preferences in the Path and Manner morpheme ordering.
The second argument against a direct cognitive-functional explanation is that it is not clear what cognitive factors can 'force' Path/Deixis verbs to occur VVC-finally.
Instead, I suppose that a combination of cognitively and diachron-ically motivated factors are responsible for the strong tendency for Path verbs to follow non-Path verbs in VVCs.
I argue that the order of components in Manner-of-Motion and Directional VVCs can be attributed to iconicity principles. At the same time, iconicity has nothing to do with the VVCs themselves, since it is rather observed in constructions underlying these VVCs. To be precise, I hypothesize that Directional VVCs and Manner-of-Motion VVCs are likely to arise from mono- and multi-clausal constructions which combine the causally-related subevents.
Consider the following VVCs from Saliba (46a-c); Eastern Kayah (47a-b) and =IHoan (48a-b). For each language, the first example is a Manner-of-Motion VVC, the second is the Directional VVC, and the third example is a Cause-Result VVC (this VVC was not in the focus of my study).
Saliba (46) a. ye-sobu-lage
3sG-dance-arrive
'He came dancing' [Margetts 1999: 119].
b. ye-tu-dobi-ei-0 3sG-throw-go.down-APP-3sG.o
'He threw it down' [Margetts 1999: 126].
c. ye-sikwa-he-beku-0 3sG-poke/hit-cAus-fall-3sG.o
'He poked it to make it fall' [Margetts 1999: 118].
Eastern Kayah
(47) a. pipe jo cwa r
butterfly fly go r0
'They (butterflies) flew away' [Solnit 1997: 77].
b. ?a de the di pld kü dir phre khu 3 put go.up at:u box in at:u shelf on
'They put it (up) in a box on a shelf' (Solnit 1997: 73).
c. ?a chw sÄ lü ne Pithoa... 3 stab die 3obv obl knife
'They stabbed him to death with a knife' [Solnit 2006: 150]. Hoan
(48) a. ma tyaqd ! ''üü sd ki !öä na
1sg walk enter obl house inside
'I walk into the house' [Berthold, Gerlach 2017: 168].
b. ma ya !ane sui...
1sg prog carry descend/lower
'I take something down.' [Berthold, Gerlach 2017: 169].
c. ma I \qäe Okoa ci 1sg pst beat kill 3pl
'I beat them dead' [Collins 2002: 56].
The semantic similarity between Manner-of-Motion, Directional and Cause-Result VVCs was acknowledged by various authors for Saliba [Margetts 1999: 117] and Eastern Kayah [Solnit 2006: 150].
Indeed, the telic Path 9 expression in the VVCs is similar to the expression of Result in Cause-Result VVCs: in (48a) the Result of 'walking' is 'entering'; in (47b), the Result of 'putting' is the 'going up' of the entity 'put'.
The fact that Cause subevents are similar to Manner subevents (see the discussion about the distinction between Manner and Cause in [Talmy 2000: 27-29]) corroborates my hypothesis that Manner-of-Motion VVCs can originate from Cause-Result constructions. Similarity between Cause and Manner can be illustrated by (48) from =IHoan VVC above. The VVC in (48c) can be understood as 'I killed them by beating' or 'I beat them, and this killed them'. Another fact to additionally support this hypothesis of 'Cause ^ Manner development' is that at least some languages have Cause-Result constructions acknowledged as a possible source of classif-icatory causative prefixes indicating the Manner of causation events; see [Bradshaw 1982: 23-24] for Papuan Tip languages and [Margetts 1999: 114-117] for Saliba.
My hypothesis is that Manner-of-Motion VVCs can develop from the sequence [Manner subevent]-[Path/Deixis-expressing Motion subevent], where the Manner subevent has Cause-like interpretation and the Motion subevent has Result-like interpretation of the type 'The butterflies flew (and) went away'^ 'The butterflies fly-went away'. Cause-Result event sequences are known to be expressed iconically; see [Durie 1997; Aikhenvald 2006]. The iconic order is thus preserved in Manner-of-Mo-tion VVCs that are not temporally sequential themselves.
A similar development pathway can be proposed for Directional VVCs. Constructions expressing Cause events and Motion events (caused by preceding events) are reinterpreted as [Action]-[Path/Orientation of a participant of the Action] complex events. Hence, I hypothesize that (46) from Saliba arose from something like 'I throw it, (it) goes down'.
9 The Path of a Motion event is telic if the boundary of the Ground is reached or crossed (e.g., in (48a)). The Path of a Motion event is atelic if the boundary of the Ground is not crossed (e.g., The girl ran toward the fence). For further discussion of Path telicity, see [Imbert 2014].
4. Discussion. Ordering principles in verb-verb compounds and parallels with serial verb constructions
To begin with, my study demonstrates that the ordering of Head and Modifier components in VVCs is not random. While the order of components is language-specific, it is still influenced by various, though heterogeneous, factors.
The order of verbs in Mental Process VVCs correlates with the order of syntactic constituents and the order of Heads and Modifiers in incorporation constructions. As for Manner-of-Motion and Directional VVCs, their component ordering is better attributed to other principles (seems to abide by other principles??). I argue that the Head and Modifier asymmetry does not influence the component ordering in Manner-of-Motion and Directional VVCs: the ordering preference in these constructions can be better explained by the distinction between Path and non-Path Motion verbs. I argue for source-oriented factors [Cristofaro 2017] as a possible explanation of the [non-Path]-[Path] component order preference in Man-ner-of-Motion and Directional VVCs. I hypothesize that the order of components in these VVCs reflects the order of constituents in less close-knit constructions from which these VVCs have (presumably) developed. This hypothesis needs more testing in further studies.
Additionally, in terms of component order, some types of VVCs are more similar to semantically corresponding less close-knit SVCs than to VVCs with different semantics. Speaking of the Path-second tendency I have observed for Manner-of-Motion and Directional VVCs (Section 3.2), I have not encountered any counterexamples in earlier SVC studies such as [Foley, Olson 1985; Givon 1991; Durie 1997; or Aikhenvald, Dixon 2006]. To highlight this point, I briefly discuss some Manner-of-Motion SVCs/ VVCs in VO languages and Directional SVCs/VVCs in OV languages.
Thai Manner-of-Motion VVCs (Thai-Kadai family, Eurasia, see [Diller 2006: 164-165]) and Tetun Dili Manner-of-Motion SVCs (Aus-troasiatic family, Eurasia, see [Hajek 2006: 242-244]) exhibit the [Man-ner]-[Path] component order. Both languages exhibit the VO dominant order, see [Diller 2006; Hajek 2006].
Taiwan Sign Language 10 employs several means of Motion event expression [Tai, Su 2013]. When a Manner and a Path verb combine in an SVC, the Manner verb precedes the Path verb as in (49) [Tai, Su 2013: 90-93]. It is worth noting that Taiwan Sign Language displays a VO dominant order [Smith 2005: 197].
(49) FROG CRAWL JARpro+FROGpro-move.out 11
Figure Motion.Manner Ground.pro+Figure.pro-Motion.Path 'The frog crawled out of the jar' [Tai, Su 2013: 93].
Path and Deictic verbs also tend to follow non-Path verbs in Directional SVCs in other languages than those in my sample such as Tariana (Arawak language, South America), see [Aikhenvald 2006: 185-188]. The Khwe language 12 (Khoe-Kwadi family, Africa) shows the same order of components in Directional VVCs, see [Kilian-Hatz 2006: 115-116]. Both Tariana and Khwe exhibit the OV dominant order, see [Aikhenvald 2003; Kilian-Hatz 2006].
Thus, the distinction between 'one-word' SVCs (VVCs) and 'multiword' SVCs may not be directly manifested in their syntax: given that different factors of both cognitive and diachronic, source-oriented nature shape the morpho-syntax of different semantic types of SVCs, it is the semantics and functions of a particular type of SVC that shape its syntax, rather than the mere level of morphological integration of SVC components.
10 I especially wish to thank Elena Alexandrovna Pasal'skaya for information about Taiwan Sign Language, Motion in Sign Languages in general and the fruitful discussion of these and many other issues.
11 A note about transcription should be made. Tai, Su [2013] use the abbreviation pro for 'proform' (a replaced form for preceding argument). The + sign indicates that two components are signed simultaneously by both hands. The — sign means that the proform(s) and the Motion / Path verb are expressed as a unit by moving hand. (a hand movement??)
12 This language name is provided in [Kilian-Hatz 2006]. I could not find this name in Glottolog.
Abbreviations
For Hup
1 — 1st person; 2 — 2nd person; 3 — 3rd person; art — Article; cf—Constant feature; edo — External direct obj; hab — Habitual; hh — Hither; imp — Imperative; ind — Indicative; loc — Locative; neg — Negation; ns—Non-singular; p — Plural; part—Particle; po—Primary obj; pfps—Perfect persistent; pvn—Perfect verb noun; ref—Reflexive; s — Subject/single argument; sfr — Stem formative; th—Thither; tu—back
For Mapudungun
1 — 1st person; 2 — 2nd person; 3 — 3rd person; art — Article; cf—Constant feature; edo — External direct obj; hab — Habitual; hh — Hither; imp — Imperative; ind — Indicative; loc — Locative; neg — Negation; ns—Non-singular; p — Plural; part—Particle; po—Primary obj; pfps—Perfect persistent; pvn—Perfect verb noun; ref—Reflexive; s — Subject/single argument; sfr — Stem formative; th—Thither; tu—Back
For Ese Ejja
1 — 1st person; 3 — 3rd person; a —Agent-like argument; abs —Absolutive; all — Allative; ds—Different subject; erg—Ergative; gen—Genitive; npf—Noun prefix; pas—Past; perl—Perlative; pl—Plural; prs—Present; res—Resultative; red — Reduplication; rpas—Remote past; sg — Singular; ss—Same subject; tel—Telic; tmp — Temporal subordinate; tr—Transitive
For Tlachichilco Tepehua
1 — 1st person; 3 — 3rd person; ald — Already; ap —Antipassive; art—Article; dat — Dative applicative; dir—Directive; dist — Distal; fut—Future; imp-fv—Imperfective; irrr—Irrealis; neg — Negation; obj—Object; pfv—Perfective; rep — Repetitive; pl — Plural; poss—Possessive; prep — Preposition; sub—Subject
For Chimalapa Zoque
1 — 1st person; 1>2 — 1st person Subject, 2nd person Object; 2—2nd person; 3 — 3rd person; a —Absolutive; antip —Antipassive; appl —Applicative; caus—Causative; com — Completive; D—Dependent; def—Definite article; e — Ergative; impv— Imperative; inc — Incompletive; N—Negative aspect; neg—Negative; nom3 — Nominalizer; npl—Plural marker (nominal); npl2—Plural marker (pronominal); pl—Plural; prn—Pronoun; rel—Relative; repet—Repetitive; vers—Versive
References
Aikhenvald 1999 — A. Y. Aikhenvald. Serial constructions and verb compounding evidence from Tariana (North Arawak). Studies in Language. International Journal sponsored by the Foundation "Foundations ofLanguage"'. 1999. Vol. 23. Iss. 3. P. 469-497. DOI: 10.1075/sl.23.3.02aik.
Aikhenvald 2003 — A. Y. Aikhenvald. A grammar of Tariana, from northwest Amazonia. Cambridge: Cambridge University Press, 2003.
Aikhenvald 2006 — A. Y. Aikhenvald. Serial verb constructions in Tariana. A. Y. Aikhenvald, R. M. W. Dixon (eds.). Serial Verb Constructions: A Cross-Linguistic Typology. Oxford: OUP, 2006. Vol. 2. P. 178-200.
Aikhenvald 2006 —A. Y. Aikhenvald. Serial verb constructions in typological perspective. A. Y. Aikhenvald, R. M. W. Dixon (eds.). Serial Verb Constructions: A Cross-Linguistic Typology. Oxford: OUP, 2006. Vol. 2. P. 1-68. DOI: 10.1093/ oso/9780199279159.003.0001.
Aikhenvald, Dixon 2006 —A. Y. Aikhenvald, R. M. W. Dixon (eds.). Serial Verb Constructions: A Cross-Linguistic Typology. Vol. 2. Oxford: OUP, 2006.
Baker et al. 2005 — M. C. Baker, R. Aranovich, L. Golluscio. Two types of syntactic noun incorporation: Noun incorporation in Mapudungun and its typological implications. Language. 2005. Vol. 81. No. 1. P. 138-176. DOI:10.1353/lan.2005.0003.
Berthold, Gerlach 2017—F. Berthold, L. Gerlach. Serial verb constructions in N!aqri-axe. Khoisan Languages and Linguistics: Proceedings of the 4th International Symposium July 10-14, 2011, Riezlern/Kleinwalsertal. Köln: Rüdiger Köp-pe Verlag, 2017.
Bickel, Züniga 2017—B. Bickel, F. Züniga. The 'Word' in Polysynthetic Languages: Phonological and Syntactic Challenges. M. Fortescue, M. Mithun, N. Evans (eds.). The Oxford Handbook of Polysynthesis. Oxford: OUP, 2017. P. 158-185. DOI: 10.1093/oxfordhb/9780199683208.013.52.
Bickel, Nichols 2007—B. Bickel, J. Nichols. Inflectional morphology. T. Shopen (ed.). Language typology and syntactic description. Cambridge: Cambridge University Press, 2007. Vol. 3. P. 169-240.
Bohnemeyer et al. 2007—J. Bohnemeyer, N. J. Enfield, J. Essegbey, I. Ibarretxe-An-tunano, S. Kita, F. Lüpke, F. K. Ameka. Principles of event segmentation in language: The case of motion events. Language. 2007. Vol. 83. No. 3. P. 495-532.
Bradshaw 1982 — M. J. Bradshaw. Word order change in Papua New Guinea Austro-nesian languages. Doctoral thesis. Hawaii: University of Hawaii, 1982.
Bril 2004—I. Bril. Complex nuclei in Oceanic languages: Contribution to an areal typology. I. Bril, F. Ozane (eds.). Complex predicates in Oceanic languages:
Studies in the dynamics of binding and boundness. Berlin, Boston: De Gruyter Mouton, 2004. P. 1-48.
Campbell et al. 1986—L. Campbell, T. Kaufman, T. C. Smith-Stark. Meso-America as a linguistic area. Language. 1986. Vol. 62. No. 3. P. 530-570.
Caballero et al. 2008 — G. Caballero, M. J. Houser, N. Marcus, T. McFarland, A. Py-cha, M. Toosarvandani, J. Nichols. Nonsyntactic ordering effects in noun incorporation. Linguistic Typology. 2008. Vol. 12. No. 3. P. 383-421. DOI: 10.1515/ LITY.2008.042.
Collins 2002 — C. Collins. Multiple verb movement in |Hoan. Linguistic Inquiry. 2002. Vol. 33. No. 1. P. 1-29.
Collins, Gruber 2014 — C. Collins, J. S. Gruber. A Grammar of \Höä. Köln: Rüdiger Köppe Verlag, 2014.
Comrie 1980—B. Comrie. Morphology and word order reconstruction: Problems and prospects. J. Fisiak (ed.). Historical morphology. Berlin; New York: De Gruyter Mouton, 1980. P. 83-96. DOI: 10.1515/9783110823127.83.
Cristofaro 2017 — S. Cristofaro. Implicational universals and dependencies. N. J. Enfield (ed.). Dependencies in language: On the causal ontology of linguistic systems. Berlin: Language Science Press, 2017. No. 2. P. 9-22.
Croft 2001 — W. Croft. Radical construction grammar: Syntactic theory in typological perspective. Oxford: Oxford University Press, 2001.
Diller 2006 — A. V. Diller. Thai serial verbs: Cohesion and culture. A. Y. Aikhen-vald, R. M. W. Dixon (eds.). Serial Verb Constructions: A Cross-Linguistic Typology. Oxford: OUP, 2006. Ch. 7. P. 160-177. DOI: 10.1093/ oso/9780199279159.003.0007.
Durie 1997 — M. Durie. Grammatical structures in verb serialization. A. Alsina, J. Bresnan, P. Sells (eds.). Complex predicates. Stanford: Center for the Study of Language and Information, 1997. P. 289-354.
Dunn 1999—M. Dunn. A Grammar of Chukchi. PhD thesis. Canberra: Australian National University, 1999. DOI: 10.25911/5D77842288837.
Dryer 2013 — M. S. Dryer. Determining Dominant Word Order. M. S. Dryer, M. Haspelmath (eds.). The World Atlas of Language Structures Online. Leipzig: Max Planck Institute for Evolutionary Anthropology, 2013. Available at: http:// wals.info/chapter/s6 (accessed on 24.05.2018).
Enfield 2009 — N. J. Enfield. Review of the book Serial verb constructions: A cross-linguistic typology ed. by Alexandra Y. Aikhenvald and R MW Dixon. Language. 2009. Vol. 85. P. 445-451. DOI: 10.1353/lan.0.0124.
Epps 2008 — P. Epps. A grammar of Hup. Berlin; New York: De Gruyter Mouton, 2008.
Evans 2003 — N. Evans. Bininj Gun-wok: a pan-dialectal grammar of Mayali, Kunwinjku and Kune. (Pacific Linguistics 541). Canberra: Pacific Linguistics;
Research School of Pacific and Asian Studies; Australian National University, 2003.
Evans 2017—N. Evans. Polysynthesis in Dalabon. M. Fortescue, M. Mithun, N. Evans (eds.). The Oxford Handbook of Polysynthesis. Oxford: OUP, 2017. DOI: 10.1093/oxfordhb/9780199683208.013.43.
Foley 1991 — W. A. Foley. The Yimas language of New Guinea. Stanford: Stanford University Press, 1991.
Foley, Olson 1985 — W. A. Foley, M. Olson. Clausehood and verb serialization. J. Nichols (ed.). Grammar inside and outside the clause. Cambridge: Cambridge University Press, 1985. P. 17-60.
Gaeta 2008—L. Gaeta. Constituent order in compounds and syntax: typology and dia-chrony. Morphology. 2008. Vol. 18. P. 117-141. DOI: 10.1007/s11525-009-9125-x.
Givön 1991 — T. Givön. Serial verbs and the mental reality of 'event': Grammatical vs. cognitive packaging. E. C. Traugott, B. Heine (eds.). Approaches to gram-maticalization. Amsterdam; Philadelphia: John Benjamins Publishing Company, 1991. Vol. I. P. 81-127. DOI:10.1075/tsl.19.1.06giv.
Givön 2009—T. Givön. Multiple routes to clause union: The diachrony of complex verb phrases. T. Givön, M. Shibatani (eds.). Syntactic complexity: Diachrony, acquisition, neuro-cognition, evolution. Amsterdam; Philadelphia: John Benjamins Publishing Company, 2009. P. 81-118. DOI: 10.1075/tsl.85.04mul.
Hajek 2005 — J. Hajek. Serial verbs in Tetun Dili. A. Y. Aikhenvald, R. M. W. Dixon (eds.). Serial Verb Constructions: A Cross-Linguistic Typology. Oxford: OUP, 2005. P. 239-253. DOI: 10.1093/oso/9780199279159.003.0011.
Hammarström, Donohue 2014—H. Hammarström, M. Donohue. Some principles on the use of macro-areas in typological comparison. Language Dynamics and Change. 2014. Vol. 4. Iss. 1. P. 167-187. DOI: 10.1163/22105832-00401001.
Haspelmath 2010—M. Haspelmath. Comparative concepts and descriptive categories in crosslinguistic studies. Language. 2010. Vol. 86. No. 3. P. 663-687.
Haspelmath 2011 — M. Haspelmath. The indeterminacy of word segmentation and the nature of morphology and syntax. Folia linguistica. 2011. Vol. 45. No. 1. P. 3180. DOI: 10.1515/flin.2011.002.
Haspelmath 2016—M. Haspelmath. The serial verb construction: Comparative concept and cross-linguistic generalizations. Language and Linguistics. 2016. Vol. 17. Pt. 3. P. 291-319. DOI: 10.1177/2397002215626895.
Imbert 2013 — C. Imbert. Morpheme Order Constraints Upside Down: Verticality and Other Directions. Annual Meeting of the Berkeley Linguistics Society. 2013. Vol. 39. Pt. 1. P. 396-411. DOI: 10.3765/bls.v39i1.3895.
Johnson 2000—H. A. Johnson. A grammar of San Miguel Chimalapa Zoque. Doctoral thesis. Austin: University of Texas, 2000.
Kilian-Hatz 2005 — C. Kilian-Hatz. Serial verb constructions in Khwe (Cen-tral-Khoisan). A. Y. Aikhenvald, R. M. W. Dixon (eds.). Serial Verb Constructions: A Cross-Linguistic Typology. Vol. 2. Oxford: OUP, 2006. P. 108-123. DOI: 10.1093/oso/9780199279159.003.0004.
Kung 2007 — S. S. Kung. A descriptive grammar of Huehuetla Tepehua. PhD thesis. Austin: University of Texas, 2007.
Li, Thompson 1973 — C. N. Li, S. A. Thompson. Serial Verb Constructions in Mandarin Chinese: Coordination or Subordination? Papers from the Comparative Syntax Festival, Chicago Linguistics Society. 1973. Vol. 9. P. 96-103.
Lieber, Stekauer 2009—R. Lieber, P. Stekauer (eds.). The Oxford handbook of compounding. Oxford: OUP, 2009.
Margetts 1999 — A. Margetts. Valence and transitivity in Saliba: An Oceanic language of Papua New Guinea. PhD thesis. Nijmegen: Radboud University Ni-jmegen, 1999. DOI: 10.17617/2.2057646.
McKay 1975 — G. R. McKay. Rembarrnga: a language of central Arnhem land. PhD thesis. Canberra: Australian National University, 1975.
Mithun 1984 — M. Mithun. The evolution of noun incorporation. Language. 1984. Vol. 60. No. 4. P. 847-894. DOI: 10.2307/413800.
Mithun 1986—M. Mithun. On the nature of noun incorporation. Language. 1986. Vol. 62. No. 1. P. 32-37. DOI: 10.2307/415599.
Muravyova et al. 2001 — I. A. Muravyova, M. A. Daniel, T. J. Zhdanova. Chukchi language and folklore in texts collected by V. G. Bogoraz. Moscow: A RSS/OS-SF report (Research Support Scheme of the Open Society Support Foundation, grant No 584/1999). 2001.
Osborne 1974 — C. R. Osborne. The Tiwi Language. Australian Aboriginal Studies. No. 55. (Linguistic Series. No. 21). Canberra: Australian Institute of Aboriginal Studies, 1974.
Panova 2017 — N. Panova. Morfologicheski svyazannyye konstruktsii s sententsial-nymi aktantami v abazinskom yazyke i v tipologicheskoyperspektive [Morphologically bound constructions expressing sentential arguments in Abaza and in a typological perspective]. Unpublished term paper. HSE, 2017.
Rose 2009 — F. Rose. The origin of serialization: The case of Emerillon. Studies in Language. 2009. Vol. 33. No. 3. P. 644-684. DOI: 10.1075/sl.33.3.05ros.
Slobin 2004 — D. I. Slobin. The many ways to search for a frog: Linguistic Typology and the Expression of Motion Events. S. Stromqvist, L. Verhoeven (eds.). Relating events in narrative. Vol. 2. Typological and contextual perspectives. Mah-wah: Lawrence Erlbaum Associates Publishers, 2004. P. 219-257.
Solnit 1997—D. B. Solnit. Eastern Kayah Li: grammar, texts, glossary. Honolulu: University of Hawaii Press, 1997.
Solnit 2006 — D. B. Solnit. Verb serialization in Eastern Kayah Li. A. Y. Aikhenvald, R. M. W. Dixon (eds.). Serial Verb Constructions: A Cross-Linguistic Typology. Vol. 2. Oxford: OUP, 2006. P. 144-159. DOI: 10.1093/ oso/9780199279159.003.0006.
Seiler 1985 — W. Seiler. Imonda, a Papuan language. Canberra: The Australian National University, 1985. DOI: 10.21256/zhaw-4756.
Smeets 2008—I. Smeets. A grammar ofMapuche. Berlin, Boston: De Gruyter Mouton, 2008. DOI: 10.1515/9783110211795.
Smith 2005 — W. H. Smith. Taiwan Sign Language research: an historical overview. Language and Linguistics. 2005. Vol. 6. P. 187-215.
Stewart 1963 — J. M. Stewart. Some restrictions on objects in Twi. Journal of African Languages. 1963. Vol. 2. Pt. 2. P. 145-149.
Tai, Su 2013 — J. H. Tai, S. Su. Encoding motion events in Taiwan Sign Language and Mandarin Chinese: Some typological implications. G. Cao, H. Chappell, R. Djamouri, T. Wiebusch (eds.). Breaking down the barriers: interdisciplinary studies in Chinese linguistics and beyond. Taipei: Institute of Linguistics; Academia Sinica, 2013. P. 79-98.
Talmy 2000 — L. Talmy. Toward a cognitive semantics. Vol. II. Cambridge, MA: MIT Press, 2000.
Voort 2004—H. van der Voort. A grammar of Kwaza. Berlin; New York: De Gruyter Mouton, 2004. DOI: 10.1515/9783110197280.
Vuillermet 2012 — M. Vuillermet. A grammar of Ese Ejja, a Takanan language of the Bolivian Amazon. PhD thesis. Lyon: Université Lumière Lyon 2, 2012.
Vuillermet 2017—M. Vuillermet. Verb compounding in Ese'eja (Takanan). Amerindia. 2017. Vol. 39. No. 1. P. 175-210.
Watkins 1984 — L. J. Watkins. A Grammar of Kiowa. Lincoln: University of Nebraska Press, 1984.
Watters 1988—J. Watters. Topics in Tepehua grammar. PhD thesis. Berkeley: University of California, 1988.
Wojdak 2005 — R. Wojdak. The linearization of affixes: Evidence from Nuu-chah-nulth. Doctoral thesis. Vancouver: University of British Columbia Library, 2005.
Züniga 2006—F. Züniga. Mapudungun. (Languages of the World / Materials, 376). München: Lincom, 2006.
Züniga 2017—F. Züniga. Mapudungun. M. Fortescue, M. Mithun, N. Evans (eds.). The Oxford Handbook of Polysynthesis. Oxford: OUP, 2017. P. 696-720.
Zwicky 1985 —A. M. Zwicky. Heads. Journal of linguistics. 1985. Vol. 21. No. 1. P. 1-29.
Получено / received 15.05.2023
Принято/ accepted 03.07.2023
