[Review of:] A. Behnke, B. Wagner-Nagy (eds.). Clause linkage in the languages of the Ob-Yenisei area: Asyndetic constructions. Leiden; Boston: Brill, 2024.

Full Text

Clause linkage is a domain of syntax that does not often receive much attention in reference grammars. It often happens that in a short 200-pages grammar more than half of the book is dedicated to morphology and grammatical semantics and less than 30 pages to syntax, where subordination is represented by singular examples. As far as Uralic and Altaic languages of Russia are concerned, this problem is reinforced by the fact that traditional grammars written in the USSR times and earlier were oriented towards the description of lexical subordinators and conjunctions, while non-finite subordination was scarcely mentioned in chapters devoted to verbal non-finite forms. It is only in the 1980s when the Novosibirsk linguistic school elaborated a model of non-finite subordination [Cheremisina 1980; Ubryatova, Litvin 1986] that linguists started to systematically consider non-finite clauses as a subtype of subordination. Given that in the Uralic and Altaic languages non-finite subordination is the only or at least the prevailing model (though in some of them it has been recently replaced by finite subordination with lexical subordinators), such an approach resulted in inadequate description of subordination as a whole.

In light of that, the volume on clause linkage in the languages of the Ob-Yenisei area is highly welcome, since it provides a large amount of new interesting data, a great deal of which has been previously ignored by the typological and theoretical literature. A noteworthy feature of the book is the use of the unified format of language description throughout the volume for every examined language. Each chapter is organized along a specific scheme and provides, first, information about the language, second, details about non-finite verbal forms, and third, a sketch on clause linkage, including coordination, adverbial subordination, relative clauses, and complementation. The book comprises sketches of 9 languages: Enets, Tundra Nenets, Evenki, Ket, Eastern Khanty, Mansi, Selkup, Chulym Turkic, and Kamas (some of them represented by two or more varieties).

The languages under discussion show a number of common properties in the domain of clause linkage, that is:

— extensive use of juxtaposition of clauses as a means of coordination and for some typically subordinate semantic relations, such as reason (Ob-Ugric, Selkup), purpose (Ket, Evenki), time (Khanty, Selkup), condition (Mansi, Selkup, Chulym Turkic), and complementation with utterance and mental verbs;

— clause chaining (Chulym Turkic, Evenki, and Southern Samoyedic);

— converbs (all the languages except for Ket);

— specific forms encoding purpose and dependent clauses as complements of manipulation and speech causation verbs, i.e. infinitives and/or supines (all the languages except for Ket, see the arguments against the term “infinitive” for the Ket action nominal in Andrey Nefedov’s chapter; in Evenki, Enets, and Nenets the general converb is used in this function, which may be termed as “infinitive”, see the remarks in the corresponding chapters);

— use of possessive markers to signal agreement with the subject of the dependent clause, at least in some subordination types (all the languages studied in the volume, with a reservation that in Chulym Turkic the possessive occurs on the head noun of a relative clause; Ket is an exception, since it lacks the nominal possessive paradigm, which is found in the neighboring languages; some subordinators, though, require for the possessive augment, whose function is unclear, see footnote 14 on p. 221);

— use of participles in relative clauses (all the languages, except for Ket).

An interesting feature of the languages under discussion is the fact that, when differentiating between same-subject and different-subject constructions, possessed subjects, especially bodyparts and some other objects of possession, may be treated in the same way as the same subjects. For example, Enets manner clauses with the general converb are mostly same-subject; however, the bodypart of the subject of the matrix clause is also possible in the subject position in a converb clause:

(1) Enets

seta oðie teʔinaad’uʔ, teinoʔ baðotia.

[se-ta ɔzi-e teʔi-naad’u-ʔ],

face-obl.poss3sg be.seen(ipfv)-cvb up-adv-lat.adv

teino-ʔ bazotia

there-lat.adv lie(ipfv).3sg.s

‘She is lying there face up.’ [Shluinsky, Wagner-Nagy 2024: 69]

Another remarkable feature of the languages under examination is the possibility to combine wh-morphemes with non-finite verbal forms in one dependent clause. This is largely observed in indirect questions in Enets (219), (220), Evenki (30c), Eastern Khanty (20), (42a) (see [Serdobolskaya et al. 2012] on similar constructions in Finno-Ugric languages).

Some languages of the Ob-Yenisey area demonstrate headless relative clauses (Enets, Eastern Khanty, Ket, Chulym Turkic), correlative constructions (Enets, Eastern Khanty, Evenki (30b), Chulym Turkic (28), Selkup). The authors explain the use of the correlatives by Russian influence; however, it must be noted that Standard Russian lacks definite correlative constructions of the who… that type (only constructions with universal reading are possible, Kto ne rabotaet, tot ne est ‘The one who does not work, does not eat’; see [Belyaev, Haug 2020]). Hence, such generalizations should be tested based on the data of neighboring Russian dialects.

Many languages under consideration have a rich and elaborated domain of temporal adverbial clauses and purpose clauses, while adversative, contrastive, and reason clauses may not receive a specific encoding, and the corresponding relations are often inferred from the context and a specific linear ordering of the clauses. With reason clauses, this pattern is observed in Enets, Tundra Nenets, Evenki, and Eastern Khanty. Adversative relations may also be encoded by conjunctions borrowed from Russian, while Tundra Nenets, Selkup, Khanty, and Mansi have native adversative conjunctions. The contrastive relation is not encoded by native conjunctions in either of the languages in question.

The terminology that is generally used for non-finite forms by most linguists presumes that, in a given language, they can be clearly differentiated on the basis of subordinate clauses they introduce. That is, converbs are expected to introduce adverbial clauses, participles are used in relative clauses, and infinitives and nominalizations head complement clauses. This idealistic picture, however, finds its exceptions in most cases. Especially Uralic and Altaic languages have a large number of non-finite verbal forms that may be assigned two or even three different labels, even if we only consider their functions. In this respect, the descriptions offered in the book under discussion offer valuable comparison of terminological labels used in different grammars and provide important discussions of the terminology. See, for example, overviews of participles, action nouns, and converbs in Nikolett Mus’s chapter on Tundra Nenets. However, we cannot fully adopt the motivation behind the choice of some terms in this and some other chapters. For example, when discussing a form that shows the properties of both infinitives and converbs, the author uses a functional motivation only: “Additionally, certain verbs and adjectival predicates can take converbs as their complements (either as a subject or as an object) in Tundra Nenets ⟨…⟩. This might lead us to categorize them as infinitives, i.e., they appear as clausal complements just like infinitives do” (p. 142). However, infinitives are termed as such on the basis of both their function and morphosyntactic properties: lack of person/number agreement, tense and mood distinction, sometimes also voice and aspectual distinctions, see [Iljevskaya, Kalinina 2002]. Therefore, a functional criterion is not enough to call a verbal form an infinitive.

Another weak point of many existing grammars, in what concerns clause linkage, is the fact that many authors hardly ever give examples of different-subject constructions. However, it is crucial for description, because many languages make use of specific markers for same-subject structures (e.g. Russian converbs), while other markers can be used irregardless of the same-subject vs. different-subject differentiation. The chapters in the book under consideration analyze all verbal forms in detail and consistently comment on the possibility of expressing the subject different from that of the matrix clause overtly, as well as the possibility to express other arguments and adverbials. This is another problematic point in many reference grammars: examples of “full” non-finite clauses, that is, clauses containing all the arguments and adjuncts of the non-finite predicate are often absent from the grammars. Sometimes it is accidental; however, in some cases it is a result of the lexicalization of considered forms, which do not introduce full clauses, but tend to be used to refer to a situation without expressing its participants. An example is found in Shoksha Mordvin, where participles are hardly ever used in relative clauses and may be termed “deverbal adjectives” (Aralova, Brykina, p. c.).

An important point is the differentiation of purpose constructions with motion verbs and purpose constructions with other types of matrix verbs, which is made in the volume, see the chapter on Enets (p. 70), in Ket (p. 226), Mansi (pp. 322 and 338), Kamas (p. 436). These constructions often demonstrate different morhosyntactic properties, whereas the contexts of function, e.g. “a rag for cleaning the floor”, and that of “give” often behave in the same way as that of motion verbs.

All these issues are covered in the book under review, thus offering valuable information for typological generalizations.

A similar problem occurs in the description of coordination: when illustrating the use of coordinating conjunctions, many grammars only provide examples of phrases (John and Mary, here and there), and the reader remains uninformed about the ability of certain conjunctions to coordinate clauses. It must be said that in some chapters of the discussed book, coordination, especially conjunction, is illustrated mostly by examples of coordinated verbs (or other types of phrases), while examples of whole clauses are scarce. Meanwhile, in some languages it is crucial, since conjunctions coordinating singular verbs or verb phrases (“He arrived, opened the door and started working”) are different from those that coordinate clauses (“The sun is up and the birds are singing”). In a book dedicated to clause linkage more examples of coordinated clauses would be expected.

I should make a special mention of the chapter on Enets by Andrey Shluinsky and Beáta Wagner-Nagy, which offers a meticulous and fine-grained description and contains many valuable remarks revealing the morphosyntactic properties of subordinate clauses and their functions. All the types of subordinate clauses are characterized at length; each type is covered in tables providing information on relevant morphosyntactic parameters. Enets is rich in purpose constructions, which receive a minute description, and the authors give a meticulous examination of complement-taking verbs and the constructions they govern. It is important that the authors clearly show the difference between mere juxtaposition and the direct speech construction (based on interpretation of indexicals) with utterance verbs, which is not examined in full in other chapters. All the generalizations are carefully formulated separately for Forest Enets and Tundra Enets.

The book opens with an introduction written by Anja Behnke and Beáta Wagner-Nagy, which gives background information on clause linkage and an overview on morphosyntax, the functions and use of non-finite forms in various types of complex sentences in the discussed languages. The introduction contains interesting generalizations, in particular on different types of converbs. The authors propose to distinguish between converbs with verbal and nominal properties. The “verbal” converbs do not take possessive and case markers; these are attested in Khanty, Selkup, Kamas, and Chulym Turkic. Converbs with nominal properties are found in Northern Samoyedic languages (Enets, Nenets), Evenki, and Mansi. It could be further suggested to differentiate between converbs that regularly introduce full clauses with all the verbal arguments and adjuncts vs. converbs that tend to be used in same-subject constructions, in periphrastic modal and aspectual constructions and as a part of a complex verb. Such forms are found in Kalmyk (converbs in -ad and -ǯə, according to [Mischenko 2009; Say 2009]) and in Eastern Mari (the converb in -ən, see [Arkhipov 2000]). The Mari converb may occasionally introduce full clauses; however, it is most often used to denote a background situation modifying the situation described by the matrix verb, which is foregrounded (e.g. “He entered the room laughing”, “He left grudging”), while full adverbial clauses are most often headed by other converbs.

However, it looks surprising that some important generalizations made in the introductory paper are scarcely followed by the authors of language sketches. For example, the differentiation of various types of converbs is hardly ever referenced to.

Although the book’s merits outweigh its drawbacks, there are some points that deserve further discussion. In a volume dedicated to a linguistic area it would be appropriate to provide a summary of contact-induced vs. native constructions in each chapter. At least, a comparison with neighboring languages suggests itself, given that all the necessary information is provided in the same book. However, such a summary is only given for Ket, Eastern Khanty, and Selkup, while other chapters merely mention Russian influence on the syntax of complex sentences and notice that particular conjunctions and subordinators are borrowed from Russian.

In discussions of complementation, it seems that some authors do not distinguish between indirect wh-questions, complementation, and sometimes also actor nominalization, cf. examples (9), (81)–(83) in the paper on Tundra Nenets, (31), (34), (38) on Evenki, (35a)–(35c) on Chulym Turkic, and (28a) on Selkup. It has been previously shown in typological studies that indirect question constructions may be formed differently from other types of complement clauses [Kahrel 1985; Morozova, Serdobolskaya 2020]; note that typological works on complementation, such as [Noonan 1985] and [Dixon, Aikhenvald 2006], do not consider indirect questions at all. In some languages indirect wh-questions may have morphosyntactic properties similar to relative clauses (Basque, Adyghe), unlike other complementation types. In languages with objective conjugation the wh-word in an indirect question is expected to trigger object agreement on the matrix verb, unlike clausal complements. Consider the following example from Tundra Nenets:

(2) Tundra Nenets

ńiśa-nd xăʔmaʔ-ma-mʔ nămda-ra-ś?

father-gen.poss2sg say.what-actn.ctmp-acc hear-2sg.sg.o-pst

‘You heard what your father said.’ [Mus 2024: 166]

The object agreement marker in this example could be attributed either to the theme of the verb ‘say’ (the words of the father), or to the complement clause as a whole. The first interpretation is more expected on general grounds, as entities are more likely to trigger agreement than situations; in any case, this question deserves more attention.

Interestingly, the languages under discussion make extensive use of direct quotation also in context of complement-taking verbs that do not involve speech, e.g. knowledge and perception verbs. However, in the book this claim is often not supported by examples showing that these verbs take the direct quotation construction, that is, examples with indexicals, exclamatives, or vocatives. For example, the paper on Eastern Khanty has persuasive examples with utterance verbs, but no examples with indexicals are given to show that cognition and perception verbs also take the same construction as a complement. For Mansi, an example intended to illustrate direct quotation (64) has the 3^rd person plural form referring to the subject of the matrix clause and, hence, showing indirect quotation.

The use of the term “action nouns” throughout the book seems infelicitous. This term has been suggested in Koptjevskaja-Tamm’s thesis [1988] and covers only (de)verbal nominals denoting action, while (de)verbal nouns denoting uncontrolled situations and states are excluded. Thus, the use of this term imposes a restriction on the deriving verbal stem, which is conceivably absent from (at least some of) the languages under discussion.

When considering nominalizations, most authors give a very detailed account of the use of possessive and case marking with the nominalized verb. Another interesting piece of information would be the use of nominal plural marking with nominalizations. As all the chapters demonstrate, case marking and the use of adpositions with nominalizations is usually a way to express various adverbial relations (in adverbial subordination) or may be governed by the matrix predicate (in complementation). Possessive marking is most often used to encode agreement with the subject of the embedded clause. Therefore, both these properties, although showing the presence of the nominal head, are functionally reanalyzed as clausal properties. As opposed to these two parameters, the use of the nominal plural usually signals to what extent the nominalized form is semantically close to non-derived nominals.

In the sections dedicated to complementation, the authors scarcely ever mention the use of objective conjugation with clausal arguments:

(3) Mansi

kāsal-as-anəl, mātāpriś juw

notice-pst-3pl.o mouse come.prs.3sg.s

‘They noticed that the mouse was coming.’ [Bíró 2024: 343]

According to the sources cited in [Ibid.], the objective conjugation marker in this example refers to the situation introduced by the complement clause. Given that many languages under consideration have objective conjugation (Tundra Nenets, Enets, Eastern Khanty, Mansi, Selkup, Ket), it is important to specify whether certain complementation strategies require/allow cross-reference by the 3^rd person objective markers on the matrix verb, see the discussion of the Mansi example (3) above and the remarks in the discussion of morphosyntax of Ket action nominals in Section 3.2 of Chapter 4 by Andrey Nefedov. In Enets, this agreement pattern is found with verbs ‘understand’ and ‘know’ shown in (192) and (194), respectively; it is absent in case of ‘think’ in (193) and ‘say’ in (190)–(191). Compare also (88) and (90) from Tundra Nenets, (76) and (74) from Eastern Khanty, (57) and (58) from Mansi. Comments on the semantic factors behind this agreement pattern would be highly welcome. It could also be interesting to see whether there are examples with plural agreement (with multiple clausal arguments), given that the 3sg marker might be used as a default agreement marker or even reanalyzed as a complementizer.

In what concerns Selkup, the distribution of the subjective and objective conjugation should definitely be commented on, since the glosses are not transparent here. It could be made more clear that the gloss “o” only signals the presence of the direct object in a clause, without indicating its person and number (for example, “1sg.o” (19c) or “3sg.o” (16b) does not refer to the 1^st or 3^rd person singular object, but rather denotes agreement with the 1^st or 3^rd person singular subject in the presence of the direct object). It is remarkable that the variation between subjective and objective conjugation in Selkup is sometimes attested with one and the same complement-taking verb, see examples (36a)–(36b) with ‘know’, and sometimes with the verbs having similar semantics, e.g. ‘think’ in (28c) and (40a)–(40b). The choice of conjugation type with various complement-taking verbs may puzzle the reader.

A remarkable feature of Uralic and Altaic languages is the presence of participles that can relativize several discontinuous positions on the Accessability Hierarchy (S > DO > IndO > Obl > Gen > OComp [Keenan, Comrie 1977]), see [Brykina, Aralova 2012]. This is also observed in the languages discussed in the book, see the examples from Chulym Turkic, where the participle in -GAn may be used to relativize direct objects and obliques such as a time adjunct below:

(4) Chulym Turkic

1. sur-ɣa-ŋ epči-ŋ

ask-gan.ptcp-poss2sg woman-poss2sg

‘the woman you have asked for’

1. tu-ɣan kün

be.born-gan.ptcp day

‘birthday’ [Däbritz, Karakoç 2024: 392]

It must be noted that the question of the availability of various relativization positions for each of the participles is not discussed in detail in many of the chapters. As far as Mansi, Kamas, and Selkup are concerned, the examples of relativization are scarce, the reason apparently being the lack of sources. The Mansi and Selkup examples only feature subject and object relativization, while for Kamas an example of adjunct relativization is given as well (“the land which I walked”).

All the comments above do not affect the overall high assessment of the book and contribute to the future studies of various topics it addresses. The volume brings up new data relevant for typology and theory of clause linkage, and the results of the nine studies carried out by the authors open new and deeper perspectives to studying coordination and subordination constructions in the languages of Siberia. The book is therefore highly welcome to the domain of language-specific studies as well as the typologically-oriented research on clause linkage.

ABBREVIATIONS

2, 3 — 2^nd, 3^rd person

abl — ablative case

acc — accusative

actn — action noun

adv — adverb

ctmp — contemporaneous

cvb — converb

gan.ptcp — participle in -GAn

gen — genitive case

ipfv — imperfective verb

lat — lative case

o — object

obl — oblique

pl — plural

poss — possessive

prs — present

pst — past

s — subject

sg — singular

About the authors

Natalia V. Serdobolskaya

Author for correspondence.
Email: serdobolskaya@gmail.com
Russian Federation, Moscow

References