This article studies Russian causal clauses employing formal contrasts that causal clauses manifest crosslinguistically. I explore two approaches that have been proposed to account for such contrasts: the coordination-subordination approach and the approach assuming that causal clauses represent different classes of adverbial clauses that differ in their degree of syntactic integration with the host clause. It is argued that the Russian data cannot be consistently accounted for unless a combination of the two approaches is applied. The analysis also reveals that the degrees of syntactic integration encoded by Russian causal clauses are different from those claimed for German causal clauses by previous research. Thus, while the degree of syntactic integration proves to be a valid parameter to compare systems of causal clauses crosslinguistically, its specific realizations may vary. Furthermore, the juxtaposition of the two approaches poses the question as to what the difference is between syntactically non-integrated adverbial clauses and coordinated clauses. I tentatively suggest that they too differ in terms of the degree of integration, but this difference belongs to the level of information packaging in discourse and not to the level of syntax.
It has been noted that causal clauses display a number of similar distinctions in different languages, both syntactic and interpretative. These may be related to the options of variable binding into a causal clause (cf. the contrast between the German causal markers weil and da in (1)), the scopal properties, the focusability of the causal clause (cf. an example from Russian in (2)), root phenomena, the availability of speech act and epistemic reading of the causal marker and so on (see Charnavel ; Rutherford  for English; Groupe Lambda-1  for French; Frey  for German; Iordanskaya ; Weiss  and Belyaev [2015b] for Russian; and Belyaev [2015a] for Ossetic).
There are at least two competing approaches to dealing with these facts. According to one of them, considered by Weiss (1989) and Belyaev (2015b) for Russian causal clauses, such distinctions are best analyzed in terms of coordination and subordination. According to the second approach, the distinctions are due to the fact that causal clauses are adverbial clauses that can be attached at different structural levels, manifesting different degrees of syntactic integration with their licensing clause (Charnavel 2019; Frey 2016, among others). Three degrees of syntactic integration correspond to three types of adverbial clauses, usually distinguished in the literature: central, peripheral and non-integrated adverbial clauses (see Endo and Haegeman 2019; Haegeman 2012, among many others). Note that, according to the common assumptions, the coordination-subordination parameter is more generic than the distinction between central, peripheral and non-integrated clauses in that the latter refers to adverbial (hence, subordinated) clauses only.
In this article, I investigate the restrictions on the use of Russian causal clauses based on the Russian National Corpus (RNC) data. I suggest that in order to account for the overall picture of such restrictions, a mix of the two approaches is required, i.e., both the coordination-subordination parameter and the distinction between central, peripheral and non-integrated adverbial clauses underlie the observed restrictions. On the one hand, as I argue, there is a coordinating causal marker in Russian that differs from other causal markers in that they introduce adverbial (hence, not coordinated) clauses. On the other hand, different adverbial causal clauses differ from each other in their potential for being central, peripheral and non-integrated, i.e., with regard to degree of syntactic integration with the licensing clause. If one considers the system of Russian causal clauses from this point of view, consistent correlations emerge between the properties of Russian causal clauses, both formal and interpretative, and their values in terms of the two parameters. Furthermore, the degrees of syntactic integration encoded by Russian causal clauses turn out to be different from those in German. While in German, according to (Frey 2016), the da-clause is a peripheral, and never a central, adverbial clause, the Russian poskol’ku-clause that seems to be the closest counterpart of da appears to be a central adverbial clause in some of its uses. A typologically particular phenomenon is the opposition of two Russian causal subordinators, potomu čto and ottogo čto, both meaning ‘because’. Both prove to introduce central adverbial clauses but differ in that the latter is licensed in a very deep position of its host clause (most probably, within the VP).
A more general question that emerges from the analysis is the difference, if any, between coordinated and non-integrated adverbial clauses. Since the latter are syntactically non-integrated, their subordinate status is partly conventional and mainly due to the fact that one and the same adverbial marker can often introduce both integrated and non-integrated adverbial clauses, as Frey’s (2016) analysis of German causal markers suggests. According to my analysis of the Russian data, coordinated clauses can be seen as less tightly integrated with their licensing clause than non-integrated adverbial clauses, while the nature of this distinction most probably lies not in syntax, but in discourse structure. Thus, the degree of integration can in fact be treated as a cover term for the coordination – subordination distinction, on the one hand, and the distinction between central, peripheral and non-integrated clauses, on the other.
The rest of the article is organized as follows. Section 2 presents five Russian causal markers (potomo čto ‘because’, poskol’ku ‘since’, tak kak ‘since, as’, ottogo čto ‘because’, ibo ‘for’) that are in the focus of the present study. In Section 3, I analyze the approach to Russian causal clauses in terms of coordination and subordination, elaborated by Belyaev (2015b). I argue that the causal markers he considers (potomo čto, poskol’ku, tak kak) cannot actually be differentiated in these terms. There is, however, a type of causal clause in Russian (the one introduced by the marker ibo) that manifests truly coordinating traits. In Section 4, I go on to analyze Russian causal clauses as adverbial clauses that differ in terms of the degree of their syntactic integration with the host clause. In doing so, I mainly follow the approach of Frey (2016) to German causal clauses. I suggest that the degree of syntactic integration is the relevant parameter for differentiating four causal markers (potomo čto, poskol’ku, tak kak, ottogo čto), including those considered by Belyaev. In Section 5, I summarize the overall picture of Russian causal clauses. Since, according to the previous sections, Russian has both coordinated and non-integrated adverbial causal clauses, I also tackle the question of what the differences are between the two. Section 6 concludes.
2 On the causal markers under examination
Five Russian causal markers are investigated in the present article: potomu čto (3), ottogo čto (4), poskol’ku (5), tak kak (6) and ibo (7). Potomu čto, poskol’ku and tak kak are by far the most frequent causal markers in Russian (see the quantitative data in Pekelis 2017). Ottogo čto and ibo are crucial for the analysis presented below since the former displays peculiar traits with respect to the degree of syntactic integration (see Section 4), while the latter is special in terms of coordination and subordination (see Section 3.3).
A brief comment is in order regarding the morphosyntactic properties of the markers in question. Both potomu čto and ottogo čto go back to a three-word combination po tomu čto/ot togo čto, where po/ot are prepositions with the causal meaning, tomu/togo are the Dative/Genitive forms of the distal demonstrative tot, and čto ‘that’ is the general subordination marker. Synchronically, potomu/ottogo and čto are separable both by pauses and other words, cf. the “split” variant of potomu čto in (8a) and the “unified” variant in (8b). The two variants differ prosodically (potomu/ottogo bears a pitch accent in case it is part of the split variant) and in terms of information structure, i.e., the causal meaning expressed by the unified potomu čto/ottogo čto tends to be focus, while that expressed by the split potomu čto/ottogo čto tends to be topic (Paducheva 1977). Syntactically, the split variant seems to be a step behind the unified one on the grammaticalization path between the original prepositional phrase and the causal conjunction. This is supported primarily by the separability of the components, which characterizes the split variant. While for reasons of space I will not point out the differences between the variants in what follows, I will mostly focus on the unified potomu čto and ottogo čto. Note, however, that any negative statement I will make with regard to these markers, i.e., a statement that says that a phenomenon X is disallowed or problematic for them, is to be understood as valid for both variants.
Tak kak is originally a manner construction consisting of a free combination of the demonstrative pronoun tak ‘thus’ and the relative pronoun kak ‘how’. As a causal marker, it has been substantially grammaticalized and synchronically its two components cannot be separated from each other either prosodically or by a punctuation mark (Belyaev 2016b: 38).
Poskol’ku is similar to tak kak in that it goes back to a two-word combination po ‘by, via’ and skol’ko ‘how much’, originally a degree construction. Synchronically, the two components are no longer separable either prosodically or by a punctuation mark or even orthographically.
Ibo, originally a combination of particles i and bo, goes back to Church Slavonic (Fasmer 1986: 113). It is the oldest of the five markers and the only one that does not originate from a combination of a head and its complement (for potomu čto, ottogo čto and poskol’ku, the original “head” is the preposition, for tak kak it is the demonstrative tak). It does not seem to be a coincidence that ibo is also the only truly coordinating causal marker in modern Russian, as suggested in Section 3.3.
3 A coordination-subordination approach to Russian causal clauses
I start with a brief survey of the approaches to coordination and subordination and the essentials of the (Belyaev 2015b) approach to Russian causal clauses in Section 2.1. In Section 2.2, I argue that treating the three most frequent Russian causal markers (potomu čto, poskol’ku and tak kak) within the latter approach fails to account for the differences between them. However, in Section 2.3, I demonstrate that a coordinating causal marker does exist in Russian, thus suggesting that the coordination-subordination parameter is relevant to the Russian system of causal clauses.
3.1 Approaches to coordination and subordination with the focus on Belyaev’s (2015b) approach
It is well known that treating coordination and subordination as a binary opposition runs into problems. A range of typological data shows that various types of complex clause structures do not clearly fall in either of these two categories (see, among others, Comrie 2008; Cristofaro 2003: 18; Haspelmath 1995; Kazenin and Testelets 2004). There are essentially two different ways in which typologists have approached the discrepancies between coordinate and subordinate properties. One way is to consider coordination and subordination as a continuum of clause linkage ranging from prototypical coordination to prototypical subordination. This is the approach taken in Weiss (1989) with regard to Russian causal clauses (see also Lehmann ), where the continuum approach is exemplified by the typologically diverse data. The main problem with the continuum idea is that the picture of polipredicative constructions it provides, though descriptively adequate, lacks consistency and predictability since any construction of any language can be located at a certain point of the continuum simply due to an arithmetic “summation” of its subordinate or coordinate traits.
Another way to deal with the coordination-subordination discrepancies is the multi-level approach, according to which the distinction between coordination and subordination remains binary but applies separately at different levels of the language structure. Within this approach, which goes back to such seminal works as Culicover and Jackendoff (1997) and Yuasa and Sadock (2002), constructions with “mixed” coordinate and subordinate properties are treated in terms of a mismatch between syntax and semantics. A serious advantage of the multi-level approach compared to the continuum approach is that the former provides a way to make predictions as to which kind of discrepancies may or may not be attested in the languages. It is to be expected, indeed, that there can be a conflict between different levels, i.e., between semantic and syntactic properties of a construction, but not within one and the same level (Belyaev 2015a: 274).
Belyaev (2015b) applies the multi-level approach to the analysis of Russian causal clauses. Therefore, his results are of primary importance for the present research. Contrary to Culicover and Jackendoff (1997), Belyaev suggests distinguishing between three levels of coordination and subordination instead of two. He relies on the Lexical Functional Grammar framework to spell out his proposal, hence the levels on which coordination and subordination are being separately defined are the constituent structure (c-structure) level, the functional structure (f-structure) level, and the semantic level. The c-structure level of coordination and subordination roughly corresponds to what Culicover and Jackendoff (1997) call the syntactic level, while the f-structure level corresponds to what they call semantics. The semantic level within Belyaev’s approach has no correspondence in Culicover and Jackendoff’s terminology.
What motivates the three-level distinction for coordination and subordination is essentially the fact that there is a cluster of semantic features according to which coordinate and subordinate clauses pattern differently, but which do not correspond to the diagnostics employed by Culicover and Jackendoff when they illustrate the notions of semantic coordination and subordination. This suggests that what Culicover and Jackendoff discuss as the semantic level, is in fact, not semantics.
A famous example of the semantic contrast between coordination and subordination are the German causal markers weil and denn. As is well known, clauses introduced by these markers differ syntactically in that weil-clauses display verb-final word order, typical of subordinate clauses in German, while denn-clauses display verb-second word order, typical of main clauses. Importantly, the two clauses also pattern differently with respect to focusing on the meaning expressed by the causal marker, i.e., putting it within the scope of an external operator (negation, modals, why-question and so on). Only weil can be subject to such focusing, cf. (9).
Thus, Belyaev (2015b) treats coordination and subordination separately at two syntactic levels and one semantic level, making the focusability of the meaning of the conjunction the basis for the coordination-subordination distinction at the semantic level. The meaning of a semantically subordinating conjunction can be focused by means of an external operator, while that of a semantically coordinating one cannot. Two alternative theoretical generalizations are considered in order to account for this semantic difference. According to one of them, coordinating conjunctions introduce conventional implicatures, while subordinating conjunctions introduce at-issue meanings. The second tentative generalization is that coordinate clauses correspond to different speech acts, which are linked by a rhetorical relation introduced by a coordinating conjunction, while a subordinating conjunction serves to link content within one and the same speech act. Being located beyond the speech acts expressed by the clauses, coordinating conjunctions operate at a higher level than ordinary predicates and subordinating conjunctions do, hence they cannot be focused by means of an operator that is located within these speech acts (I will come back to the second generalization when discussing the difference between coordinated and syntactically non-integrated adverbial clauses in Section 5.3).
For each level of coordination and subordination, a number of criteria are considered that serve to identify the status of a construction at the respective level. For the syntactic level, the criteria are mostly well known and often mentioned in the literature on coordination and subordination. These are criteria that test the linear order constraints, the possibility of extraction out of the clauses, and so on (see more on Belyaev’s criteria in Section 3.3). The semantic criteria test whether the conjunction meaning can be focused as an answer to a why-question or under the scope of negation, modals, focus particles, or other external operators. Strictly speaking, some of these criteria concern the information-structure level rather than semantics. However, they are semantic in that they serve to test the distinction between coordination and subordination at the semantic level, according to the definition of this distinction proposed by Belyaev.
In what follows, I will generally adhere to Belyaev’s approach to coordination and subordination sketched out above. Still, two caveats should be mentioned. On the one hand, Belyaev’s distinction between two different syntactic levels of coordination and subordination is, in my view, not put clearly enough and deserves further investigation. It is not to be excluded that the criteria associated with the second syntactic level (f-structure) can in fact be successfully distributed between one syntactic and one semantic level, i.e., that one syntactic and one semantic level would suffice to account for the whole range of coordination-subordination mismatches. On the other hand, the definition of semantic coordination and subordination in Belyaev’s terms seems to be a very insightful one. In fact, treating coordinated clauses as different speech acts, each with its own illocutionary force or its own information structure is not unprecedented; see similar proposals in Weiss (1989), Kobozeva (2000), and Verstraete (2005), among others. What is fundamentally new in Belyaev’s approach is the suggestion to define the semantic coordination and subordination within the multi-level view in these terms. Culicover and Jackendoff (1997), who were the first to propose the multi-level approach to coordination and subordination, have been much less explicit regarding what exactly lies behind the semantic distinction.
3.2 Russian causal clauses in terms of coordination and subordination: pros and contras
The multi-level approach to coordination and subordination, albeit insightful in its essentials, does not seem to account for the distribution of the causal clauses introduced by potomu čto, poskol’ku and tak kak (pace Belyaev [2015b], who refers to these conjunctions in order to illustrate the three-level approach). Belyaev’s analysis of potomu čto, poskol’ku and tak kak in terms of coordination and subordination is based on a range of diagnostics, each of which is assumed to be linked to a particular level of coordination and subordination. The major conclusions are summarized in Table 1.
tak kak is coordinating in semantics;
potomu čto is coordinating in syntax (more precisely, in c-structure);
in some of its uses, potomu čto is coordinating in semantics.
In the following subsections, I will try to show for each debatable point in Table 1 that the linguistic data that are assumed to support it, in fact, cannot be convincingly accounted for in terms of coordination and subordination. Note that for the time being, I will leave aside the analysis of ottogo čto and ibo, which are two more Russian causal markers introduced in Section 2. I will turn to the analysis of ibo in Section 3.3 and to the analysis of ottogo čto in Section 4.
3.2.1 Is tak kak coordinating in semantics?
In what follows, I consider one by one the arguments that according to Belyaev (2015b) give evidence in favor of the assumption that tak kak is semantically subordinating.
The term èto-focus refers to a construction as in (10), in which the semantic relation expressed by a conjunction is focused by means of the anaphoric pronoun èto ‘this’ that substitutes the main clause and refers to the pre-text. In (10), the situation ‘snow fell on the street’ is said to be simultaneous with the situation ‘snow had already melted in the city’, which runs contrary to expectations, and not with another situation and another point in time, which would be more to be expected. Only (semantically) subordinate clauses may be focused in this way.
Belyaev (2015b: 46) assumes that tak kak cannot participate in the èto-focus under any circumstances, while two other markers can, but poskol’ku is used only marginally in this construction. This description seems to be insufficient. First, a few examples with tak kak can be found on the internet (consider (11)). Second, the fact that poskol’ku is marginal within the èto-focus seems to be important and not negligible. Example (12) is the only example with poskol’ku attested in the main corpus of the RNC.
As the data in Table 2 show, potomu čto is more than a hundred times more frequent in this context (the difference is statistically significant, χ 2 (1, N = 226, 596) = 152.102, p < 0.001).
|èto||no èto||Total||% of uses with èto out of the total number of uses of potomu čto/poskol’ku|
– Focus particles
As is claimed by Belyaev (2015b: 46–47), tak kak differs from poskol’ku and potomu čto in that it cannot be focused by means of focus particles including negation or only-focus. But here again, it is worth noting that focusing is also rather problematic for poskol’ku. According to the data in Table 3, poskol’ku combines with the particle tol’ko ‘only’ more than 30 times less frequently than potomu čto (the difference is statistically significant, χ 2 (1, N = 226,596) = 1,040.389, p < 0.001).
|tol’ko||no tol’ko||Total||% of uses with tol’ko out of the total number of uses of potomu čto/poskol’ku|
Belyaev (2015b: 47–48) claims that tak kak cannot be used as an answer to a why-question, while poskol’ku can, albeit somewhat marginally. However, examples with tak kak can be found not only on the internet but also in the RNC. Consider the example in (13), taken from the multimodal corpus of the RNC:
Epistemic and illocutionary cause
As is well known, a causal relation can be interpreted on three different levels: the content domain, as in (14), the epistemic domain, as in (15), and the speech act domain, as in (16) (Sweetser 1990, among others).
Two assumptions are made in (Belyaev 2015b) in regard to this distinction. According to the first one, if a conjunction can be used epistemically or on the speech act level, this is a symptom of its being semantically coordinating. According to the second assumption, poskol’ku cannot be used as an epistemic causal marker, while tak kak can (hence, it is semantically coordinating due to the first assumption).
I disagree with both assumptions but will turn to the first one in Section 4 when discussing the approach to causal markers developed by Frey (2016). As for the second assumption, it is false since poskol’ku and tak kak pattern in the same way. Instances of the epistemic use are attested in the RNC for both of them, consider (17) and (18).
To summarize, the contrast between tak kak and poskol’ku does not seem to be pronounced enough for qualifying one of them (tak kak) as semantically coordinating and the other one (poskol’ku) as semantically subordinating.
3.2.2 Is potomu čto coordinating in syntax?
The assumption that potomu čto is syntactically coordinating is based on the linear order data (Belyaev 2015b: 42). Potomu čto generally disallows both embedding, as in (19a), and preposing, as in (19b).
I suggest, however, that these linear constraints are due to the information structure properties of potomu čto rather than to its syntactic status: a clause introduced by potomu čto can only convey new information (for more detail, see Pekelis 2017). This explains why such a clause is infelicitous in clause-initial position, since the latter is associated with topicality. This also explains why a clause introduced by potomu čto is infelicitous when embedded since embedded clauses are also associated with given and less important information. Not surprisingly, parenthetical clauses are typically clause medial (Zaliznjak 1992: 436–450). Crucially, a clause introduced by potomu čto is infelicitous in clause-final position when it conveys given information. In (20), it is clear from the context that the piece of information ‘they were going to see him off’, expressed by the causal clause, is given. Accordingly, potomu čto cannot be used instead of ottogo čto. Note that it would be particularly hard to account for this fact if one assumes that the linear order constraints are syntactic in nature and are due to the properties of potomu čto in terms of coordination and subordination.
Finally, the constraint on preposing of potomu čto is not that strict: it can be violated if the potomu čto-clause is modified by a focus particle, as in (21). This fact is also in line with a pragmatic account of the linear order restrictions rather than a syntactic one.
I conclude, therefore, that the linear order constraints are not sufficient for qualifying potomu čto as a syntactically coordinating marker.
3.2.3 Can potomu čto be coordinating in semantics?
The idea that potomu čto can be semantically coordinating is based on the observation that when it is used as an epistemic or a speech act conjunction, it loses its semantically subordinating properties, namely the capacity to be within the scope of external operators. In particular, it can no longer participate in the èto-focus (Belyaev 2015b: 54). In (22a), potomu čto introduces an eventuality-related causal clause (from here on, and following Frey (2016), I will use this term to refer to a causal relation within the content domain) and it is felicitous as a part of the éto-focus construction, while in (22b), potomu čto is used as an epistemic marker and is inappropriate within the éto-focus.
Furthermore, a clause introduced by an epistemic or a speech act potomu čto exhibits main clause phenomena (Belyaev 2015b: 56). Consider Example (23), in which ohota že is used, a special construction that is semantically close to a rhetorical question.
However, potomu čto is not the only marker whose use epistemically or as a speech act conjunction is associated with properties that Belayev assumes to be symptomatic of semantic coordination. Almost all adverbial subordinators pattern in the same way. I will illustrate this with a few examples.
Just as potomu čto, the causal marker poskol’ku and the concessive marker nesmotrja na to čto ‘despite the fact that’ stop being compatible with the èto-focus as soon as they are used epistemically or at the speech act level. In (24a), poskol’ku participates in the èto-focus and its use is eventuality-related, while in (24b), poskol’ku is used as an epistemic marker and the éto-focus construction is not available, according to (24c).
Similarly in (25a), nesmotrja na to čto is part of the èto-focus construction and its use is eventuality related, while in (25b) it is used at the speech act level and the éto-focus construction is infelicitous, as is illustrated in (25c).
Furthermore, Example (26) gives evidence to the assumption, spelled out in (Coniglio 2011) and (Frey 2016), that epistemic and speech act clauses typically exhibit main clause phenomena. In (26), the conditional subordinator esli ‘if’ is used as a speech act connector: the speaker puts the truth of ‘you still remember this’ as a condition for posing the question ‘what happened?’ to the addressee. At the same time, the esli-clause contains the parenthetical word konečno ‘of course’, which belongs to main clause phenomena in Russian according to Paducheva (1990).
Thus, the data that, according to Belyaev (2015b), prove potomu čto-clauses to be semantically coordinating, turn out to be not specific to potomu čto-clauses, but rather common to epistemic and speech act clauses in general. If one remains within Belyaev’s approach to Russian causal clauses, it is to be concluded that all epistemic and speech clauses are coordinating in semantics. This would mean, in turn, that a wide range of adverbial subordinators can be used as semantic coordinators, which seems a rather counter-intuitive conclusion. A more convincing interpretation of the facts under discussion is the one proposed by Frey (2016) (see also Haegeman 2012: 181), who argues that the epistemic and speech act clauses represent syntactic types of adverbials that are attached higher in the syntactic structure than the eventuality-related adverbials. I will come back to this proposal in Section 4.
3.2.4 Interim conclusions
My preliminary conclusions are as follows. It does not seem to be true that the formal distinctions displayed by three major causal markers in Russian (potomu čto, poskol’ku and tak kak) should be accounted for in terms of coordination and subordination. Instead, I would like to suggest in Section 4, following the approach developed by Frey (2016), that these markers are best described as adverbial clauses that differ in degree of their syntactic integration with the licensing clause. This does not mean, however, that the coordination-subordination parameter is irrelevant at all to the system of Russian causal clauses. In the following Section, I will turn to the causal clauses introduced by the marker ibo ‘for’, which prove to be both syntactically and semantically coordinated within the approach to coordination and subordination elaborated by Belyaev (2015a, 2015b).
3.3 Marker ibo as a coordinating causal conjunction in Russian
The conjunction ibo is supposed to be subordinating in the Russian grammatical tradition (Švedova 1980: 578). However, I will demonstrate in the present Section that ibo patterns as a coordinating marker according to the majority of criteria used by Belyaev, including those that test coordination and subordination in semantics.
It is worth noting that the very existence of causal coordinating conjunctions poses a theoretical question as to what the general relation is between the causal semantics and the coordination-subordination parameter. The typological data are rather contradictory here. On the one hand, causal markers that display coordinating traits are not a rarity at least in European languages (cf. the German denn or the English for). On the other hand, it has been noticed that causal clauses tend to be subordinated. According to Kazenin and Testelets (2004), the converb construction in Tsakhur exhibits syntactic properties of subordination in case the converb clause has the causal interpretation, but it turns out to be coordinated if no causal semantics are at play. This ambiguity is reflected in the properties of ibo, which, as I will show, patterns as a coordinating conjunction, but not as a canonical one.
Following Belyaev (2015b), I will distinguish between three groups of criteria, which correspond to three levels of coordination and subordination: linear order criteria (constituent structure level); Across-the-Board (ATB) extraction, gapping and scope of mood criteria (functional structure level); and a group of semantic criteria.
3.3.1 Linear order
According to the linear order criterion, subordinating markers, contrary to coordinating ones, typically allow both embedding and preposing of the clause they introduce. Ibo patterns as a coordinating marker with respect to this criterion. It disallows both preposing, as in (27b), and embedding, as in (27c).
In the Russian dependency treebank SynTagRus, which is a subcorpus within the RNC, no instances of ibo-clauses have been attested that are center-embedded or preposed to the main clause.
3.3.2 ATB, gapping, scope of mood
The criteria in the second group at first sight go contrary to the assumption that ibo is coordinating. Ibo disallows ATB extraction, i.e., an extraction in which the extracted constituent is simultaneously related to a gap in every clause, as in (28). The ban on ATB extraction is typical of subordinating markers.
Ibo disallows gapping as well, which is also deemed to be a symptom of subordination. Cf. (29).
However, the possibility of ATB seems to be linked directly to some sort of semantic symmetry of the interclausal relation, which is clearly absent from the case of the causal ibo, and only indirectly to the coordination-subordination distinction. Another conjunction that gives evidence to this assumption is the Russian tak čto ‘so’, cf. (30). Tak čto introduces result clauses, so it is as semantically asymmetric as ibo is. And exactly like ibo, it displays the majority of properties of a coordinating marker except for the possibility of ATB extraction (see Pekelis 2015 for details).
Note also that according to relatively recent proposals such as (Nunes 2001), the structure of the sentences with ATB-extracted elements is similar to the structure of the sentences with parasitic gaps, which usually occur in adverbial (hence, subordinate) clauses.
Gapping does not seem to be a reliable criterion for coordination and subordination either, at least in Russian. What is relevant to gapping in Russian is the presence of a semantic contrast between topics and/or foci of the combined clauses. Such a contrast is present in constructions with the Russian coordinating marker a, but is absent in constructions with the coordinating marker i, albeit both mean roughly ‘and’ (see Uryson 2011 for a detailed semantic analysis of a and i). Accordingly, gapping is felicitous with a, but unnatural with i:
The last criterion in this group is the scope of subjunctive mood assigned by the matrix verb. In this regard, ibo does not pattern as a subordinating marker. As shown in (32a) and (32b), ibo differs from potomu čto in that it cannot remain out of the scope of subjunctive assigned by the verb hotet’ ‘want’. However, it does not pattern with the canonical coordinating marker i ‘and’ either, since it cannot be within the scope of the subjunctive, cf. (32b) and (32c). Ibo is simply unembeddable together with its licensing clause, so the criterion yields no clear results.
Presumably, this again is due to the fact that ibo is coordinating, on the one hand, and introduces a semantically asymmetrical relation, on the other. Being coordinating, ibo disallows focusing of the causal relation, which occurs in embedding contexts like (32a), where the cause introduced by the potomu čto-clause is being implicitly compared with other eventual causes. At the same time, due to its asymmetric (causal) semantics ibo cannot avoid focusing of the relation it expresses, as semantically symmetric coordinate markers normally do. Consider Example (32c), in which the symmetric relation expressed by the coordinating and is not being focused while the and-clause is being embedded under the verb hotet’ together with its licensing clause. For space reasons, I will not expand on the details of this contrast here.
Thus, although this group of criteria apparently gives a piece of evidence in favor of the subordinating status of ibo, I suggest that in actual fact it fails to qualify ibo in terms of coordination and subordination.
3.3.3 Semantic criteria
Contrary to the previous group of criteria, the semantic criteria, which test whether the semantic relation expressed by the marker can be focused by external operators (see Section 3.1), allow qualifying ibo as a coordinating marker in a uniform way.
First, ibo can be used neither with focus particles tol’ko ‘only’ or imenno ‘exactly’, nor under negation. No instances have been attested either in the main corpus or the newspaper corpus of the RNC except for one, Example (33), which is highly stylistically marked.
Second, ibo cannot participate in the èto-focus construction, as shown in (34). No instances of ibo within the èto-focus have been attested either in the main or the newspaper corpus.
Third, ibo does not allow extraction out of its licensing clause, just as a coordinating marker would do in accordance with the Coordinate Structure Constraint (CSC). In Example (35), a yes/no-question with the focus in the main clause is felicitous with ottogo čto and is infelicitous with ibo. Note that Belyaev (2015b: 55), following Kehler (2002), argues that CSC serves to test coordination and subordination in semantics.
Instances of ibo in the RNC containing a wh-word within the licensing clause are all instances of the speech act use of ibo, as in (36). In such cases, the ibo-clause is not in the scope of the wh-word, hence, no violation of the CSC occurs.
Fourth, and finally, ibo can only very marginally introduce an answer to a why-question. Example (37), found on the internet, sounds rather awkward.
3.3.4 Ibo: interim conclusions
The results of different criteria applied to ibo in the previous subsections are summarized in Table 4. The shaded cells refer to the criteria that I deem to be unreliable for distinguishing coordination and subordination in Russian. My main conclusions are as follows: I assume ibo-clauses to be coordinated both in syntax and semantics. I suggest that one of the reasons for which the ATB, gapping and scope of mood criteria yielded doubtful or unclear results are the causal (asymmetric) semantics of ibo. These criteria are based mainly on canonical coordination and subordination cases, while the causal semantics are a strong bias towards subordination. For the same reason, ibo marginally allows patterns associated with subordinate structures (cf. its occasional use under negation or as an answer to a why-question).
|Linear order||ATB||Gapping||Scope of mood||èto-focus||Focus particles, NEG||CSC||why-question|
However, the portrait of ibo remains incomplete unless its stylistic traits are taken into consideration. Ibo is qualified as an archaic marker by a number of dictionaries of Russian (cf. Evgen’eva 1985: 626). But for present-day Russian, this does not seem to be correct, since ibo is frequent enough in the newspaper corpus of the RNC (6,322 instances, compared with 1,242 of ottogo čto). Interestingly, ibo was perceived as archaic in the Russian language as early as the first half of the 19th century according to Bulachovsky (1954: 395). What is therefore puzzling about ibo is the fact that a marker that has been perceived as archaic for two centuries has “survived” and is being productively used in modern newspaper texts. Given what was said above, one may speculate that one of the reasons why ibo is still in use is that it satisfies the need for a coordinating causal conjunction.
4 Russian causal clauses as syntactically different types of adverbial clauses
In this section, I analyze Russian causal clauses introduced by potomu čto, ottogo čto, poskol’ku and tak kak. Based on a number of syntactic and interpretative features displayed by these clauses, I argue that they are better differentiated as belonging to different classes of adverbial clauses, which differ in degree of syntactic integration with the host clause, than in terms of coordination and subordination. Ottogo čto-clauses turn out to be more tightly integrated than potomu čto-clauses, which, in turn, are more tightly integrated than both poskol’ku- and tak kak-clauses. Interestingly, the preposition ot, which is part of ottogo čto (cf. Section 2), differs from the preposition po, which is part of potomu čto, in that the former presupposes a tighter link between cause and effect than the latter (Boguslavskaja and Levontina 2004: 83). Most probably, this difference between the prepositions gave rise to the difference in terms of degree of integration that we find in the respective conjunctions.
Since I generally follow the (Frey 2016) approach when interpreting the distinction between different classes of adverbial clauses (central, peripheral and non-integrated clauses), I start with the essentials of this approach in Section 4.1. In Sections 4.2 and 4.3, I introduce my analysis of potomu čto-, poskol’ku-, tak kak- and ottogo čto-clauses. In Section 4.2 I survey a number of properties of these clauses, which, as I suggest, indicate different degrees of their syntactic integration. I go on to characterize potomu čto-, ottogo čto-, poskol’ku- and tak kak-clauses as belonging to different classes of adverbial clauses in Section 4.3.
4.1 The essentials of Frey’s (2016) approach to German causal clauses
The approach is based on the well-known distinction between three syntactic classes of adverbial clauses, namely, the central adverbial clauses (CACs), the peripheral adverbial clauses (PACs), and, in the terminology of Frey (2012), the non-integrated dependent clauses (NonICs). According to Frey’s (2016) treatment of this distinction, which is in many respects congruent to other approaches (Endo 2012; Endo and Haegeman 2019; Haegeman 2012, among others), the differences between the classes can be summarized as follows. A CAC is licensed in the standard way inside its host’s TP by the verb or one of its functional projections. A PAC is also syntactically licensed by its host, but contrary to a CAC is licensed in a very high position by the host’s Force-projection. A NonIC is not part of the syntactic structure of its associated clause; it is a “syntactic orphan”. Its licensing happens semantically by a rhetorical relation that connects it with its associated clause.
Frey (2016) considers three types of German causal clauses: da ‘since’-clauses, weil ‘because’-clauses that have the finite verb in second position (weil-V2-clause) and weil ‘because’ -clauses that have the finite verb in final position (weil-Vf-clause). According to his analysis, these types differ in terms of the above syntactic classification. A weil-Vf-clause is by default a CAC, but, if supported by indicators (i.e., words or constructions that emphasize the respective interpretation), it can occur as a PAC or a NonIC. A da-clause occurs as a PAC; if supported by indicators, it can occur as a NonIC. A weil-V2-clause necessarily occurs as a NonIC. Note that Frey treats both weil-Vf, weil-V2 and da-clauses as adverbial clauses. This means that, contrary to what Belyaev (2015b) suggests for Russian causal clauses, Frey does not propose to interpret any distributional differences between weil-Vf, weil-V2 and da-clauses in terms of coordination and subordination (As mentioned in the Introduction, the term “subordinate” is substantially conventional when applied to NonICs. See more on the relationship between NonICs and coordinated clauses in Section 5).
The examples in (38) and (39) illustrate the syntactic differences between weil-Vf-, da- and weil-V2-clauses. As follows from (38), the weil-Vf-clause, on the one hand, and da- and weil-V2-clause, on the other, pattern differently with respect to variable binding. In (38a), we observe binding of a quantified DP into a weil-Vf-clause. But the binding is possible neither into a da-clause, as in (38b), nor into a weil-V2-clause, as in (38c). This contrast is accounted for by the assumption that, being a CAC, the weil-Vf-clause is base generated in a deep position of the host clause, which is inside the c-command domain of the subject of the host. On the contrary, neither a da-clause nor a weil-V2-clause can occur as a CAC, hence they are base generated in a position that is outside of the c-command domain of the subject of the host.
While the da- and weil-V2-clauses pattern together with respect to binding, they behave differently with respect to the option of being positioned in the prefield of a German verb-second clause. Both CACs and PACs, hence both the weil-Vf- and da-clauses, can be positioned in the prefield, while NonICs, hence the weil-V2-clauses, cannot. Cf. (39).
The central idea of Frey’s (2016) approach is that there is a one-to-one mapping between the external syntax of a causal clause and its semantic type, i.e., the interpretative level of a causal clause corresponds to the degree of its integration with the licensing clause. Thus, a causal CAC is base generated inside its licensing clause in a deep position and refers to a relation between eventualities (so it can only express an eventuality-related meaning, but not, say, an epistemic meaning). A causal PAC is base generated inside its licensing clause in a high position and refers to a relation between propositions, understood as functions saying whether an eventuality is realized in certain worlds or not. A causal NonIC is a syntactic orphan and refers to a relation between speech acts.
As an illustration, consider the contrast between (40a) and (40b). In both examples, the causal relation is to be interpreted at the epistemic level. That is why (40a) is ill-formed: the weil-Vf-clause, being by default a CAC, can only refer to a relation between eventualities. However, in (40b), the epistemic interpretation is forced by the epistemic particle wohl. As a special “indicator” of the epistemic reading, wohl allows the weil-Vf-clause to be a PAC and to refer to a relation between propositions.
Without going into further details of the assumptions that underlie this reasoning, I will try to show in the following sections that the Russian data fit this analysis very well, though not exactly in the same way German causal clauses do.
4.2 Ottogo čto-, potomu čto-, poskol’ku- and tak kak-clauses: further properties
In the present section, I consider a number of further syntactic and semantic properties displayed by ottogo čto-, potomu čto-, poskol’ku- and tak kak-clauses. These properties, as I argue in Section 4.3, reveal that the causal clauses under examination represent different syntactic classes of adverbial clauses.
4.2.1 Scope of negation
Only ottogo čto-clauses may occur inside the scope of the matrix negation. Consider example (41a), in which it is denied that she falls into a trance because people die. Ottogo čto cannot be substituted here with any of the other three markers, since in this case the causal meaning is outside the scope of negation, which creates a rather contradictive meaning (41b).
(42), (43) and (44) are examples with potomu čto, poskol’ku and tak kak from the RNC, in which the causal meaning is outside the scope of negation. These examples do not allow an interpretation according to which the causal relation is inside the scope of negation, although semantically such an interpretation is not to be excluded. Example (42), for instance, gives a cause for why he was not capricious. The reading according to which it is denied that he was capricious because it was difficult for him to scream is not available for (42), although it would make sense.
4.2.2 Scope of epistemic modals
Contrary to negation, with respect to the scope of epistemic modals, all four clause types pattern identically: all of them can occur inside the scope of epistemic modals. Examples (45)–(48) illustrate this for each clause type. In (46), for instance, it is assumed that the reason why my friend writes me rarely is that he found another woman. Example (46) does not give a reason for the assumption ‘my friend writes me rarely’, as it would be if the causal meaning were outside the scope of the modal verojatno ‘probably’. In (47) and (48), the causal meaning is inside the scope of verojatno, too. In (47), it is supposed that the reason why the legislators took June 20, 1991, as the basis is that before that date Sberbank remained state owned. Example (47) does not give a reason why the assumption ‘the legislators took June 20, 1991, as the basis’ is made. Similarly in (48), the assumption concerns the reason why the residents did not open the door for a long time. (48) does not give a reason for the assumption, ‘the residents did not open the door for a long time’.
4.2.3 Epistemic use
While, as shown in the previous section, the causal relation introduced by any of four conjunctions may occur inside the scope of epistemic modals, the opposite is not true. Not all of them may occur outside the scope of an epistemic modal, or, in other words, not all of them can be used epistemically. The instances of the epistemic use of poskol’ku and tak kak were given in Section 3.2.1, cf. Examples (17) and (18)). The example in (49) shows that the epistemic use is also possible for potomu čto, while ottogo čto disallows it.
4.2.4 Speech act use
The phenomenon of speech act use shows a more sophisticated contrast between the four types of Russian causal clauses. Ottogo čto cannot be used as a speech act conjunction. It is ungrammatical in (50), where potomu čto is used at the speech act level, i.e., to introduce the reason why the speaker is performing the imperative speech act.
Potomu čto, as (50) illustrates, can be used as a speech act conjunction. However, this is the case only if the illocutionary force of the main clause is imperative (cf. berite ‘take’ in (50)). If the main clause is declarative, as in (51), or interrogative, as in (52), potomu čto cannot be used at the speech act level unless the potomu čto-clause and the question it refers to belong to different sentences, as in (53a). Example (53b), where potomu čto refers to a question within the same sentence, is ill-formed. Note also that the English translations of (51) and (52) are perfectly acceptable, hence potomu čto differs in this respect both from the English since and because.
The special status of the imperative is presumably due to the fact that the causal sentences with an imperative main clause always require a speech act interpretation, hence the imperative within these sentences serves to explicitly indicate that the causal clause is to be interpreted at the speech act level. In this respect, the imperative sentences differ from the interrogative and declarative ones, which normally allow an eventuality-related reading. This assumption is corroborated by the fact that ottogo čto cannot co-occur at all with an imperative main clause, whereas it can co-occur with an interrogative main clause. In the RNC, no instances of ottogo čto within an imperative sentence have been attested, while there are instances of ottogo čto within questions, which require an eventuality-related reading (cf. (54)).
Separating a potomu čto-clause and a question it refers to as different sentences, as in (53a), actually serves the same purpose. Due to such separation the causal relation is to be obligatorily interpreted at the speech act level, since the components of one and the same question (which the clauses in (53a) would be if the link between them were eventuality-related) cannot belong to different sentences. In (53b), with no separation, hence, no special indication of the speech act reading, this reading is not admissible.
It follows from this reasoning that while ottogo čto cannot be used at all as a speech act conjunction, potomu čto requires a special indicator that supports the speech act use to be present in the sentence. Note that the role of “indicators” is emphasized within Frey’s (2016) approach as well. According to Frey, a da-clause can occur as a NonIC only if supported by indicators.
Contrary to potomu čto, both poskol’ku and tak kak allow the speech act reading regardless of any special indicator being present in the sentence. In Examples (55) and (56), poskol’ku and tak kak occur within a question and both must be interpreted at the speech act level. Thus, neither of the markers, when used on this level, is limited to the imperative structure.
4.2.5 Relative versus absolute tense marking
One more parameter according to which Russian causal clauses differ is the possibility of the relative interpretation of the tense form they contain. A tense form within a subordinate clause is used relatively if it anchors the event temporal localization to the main event, and it is used absolutely if it anchors the event temporal localization to the speech act (Comrie 1985: 36). In this respect, again, ottogo čto-clauses are opposed to other clause types being the only ones to allow the relative tense marking. Consider the examples in (57): In (57a), the embedded predicate možet ‘can’ within the ottogo čto-clause is marked for present tense, which serves to express simultaneity with the time point in the past encoded by the main predicate zaperežival ‘got upset’. Thus, možet in (57a) is used relatively. None of the three other markers are admissible in the same context. Note that all of them are felicitous if the embedded predicate is marked for past tense, hence, used absolutely, as in (57b).
4.2.6 Ottogo čto, potomu čto, poskol’ku and tak kak: interim conclusions
The formal distinctions between the four clause types sketched out in Sections 4.2.1–4.2.5, are summarized in Table 5. One can see that potomu čto, poskol’lu and tak kak form a perfectly homogenous group with the only exception being that which concerns the speech act parameter. Ottogo čto behaves differently with respect to all parameters, excluding the scope of epistemic modals.
|Cause < mod||Cause < NEG||Rel. tense||Epistemic (cause > mod)||Speech act|
|potomu čto||+||−||−||+||+/− (only if supported by indicators)|
4.3 Ottogo čto-, potomu čto-, poskol’ku- and tak kak-clauses as different types of adverbial clauses
Based on the facts introduced in Section 4.2, in the present Section I characterize the ottogo čto-, potomu čto-, poskol’ku- and tak kak-clauses with respect to CAC-, PAC- or NonIC-classes. I argue that the formal distinctions observed for the four clause types are due to different degrees of syntactic integration associated with CACs, PACs and NonICs. In doing so, I reject the analysis of these clauses and of the contrasts they display in terms of coordination and subordination, proposed by Belyaev (2015b) (see also Weiss 1989). Once ottogo čto-, potomu čto-, poskol’ku- and tak kak-clauses are recognized as adverbial clauses, they cannot simultaneously be coordinated, with adverbial clauses being a subclass of subordinate clauses according to the common view.
Summarizing what is generally assumed within cartographic syntax (see in particular Cinque (1999: 106, 124); Krifka (2013: 5); Rizzi (1997: 297)), the hierarchical structure of the left periphery of a clause can be presented as in (58). Importantly, for my analysis of Russian causal clauses, the Force projection is at the very top of the clause structure, while the epistemic modality is supposed to be higher than negation, which, in turn, is higher than VP.
If one roughly accepts the structure in (58), the distributional data observed for ottogo čto-, potomu čto-, poskol’ku- and tak kak-clauses can be accounted for by the following assumptions. An ottogo čto-clause is a CAC – it can be neither a PAC nor a NonIC. It is integrated with its licensing clause somewhere at the VP level or at least lower than negation. Based on the structure in (58), these suggestions allow to account for the facts from (i) to (iv):
an ottogo čto-clause may occur inside the scope of matrix negation;
it can be used neither epistemically nor at the speech act level;
it may occur inside the scope of epistemic modals;
it allows the relative tense marking of its predicate.
The points in (i) and (iii) follow from (58) and my assumptions straightforwardly; (ii) follows from the idea, developed by Frey (2016), that the external syntax of a causal clause maps its semantic type so that a causal CAC can only refer to a causal relation between eventualities. With respect to (iv) note that complement clauses, contrary to adverbial clauses, tend to have relative tense marking (Šnittke 2020), on the one hand, and are generated inside the VP, on the other. The assumption that the ottogo čto-clause is generated somewhere at the VP level is thus perfectly in line with the fact that it allows relative tense marking.
I suggest furthermore that a potomu čto-clause is by default a CAC or a PAC; if supported by indicators, it can be a NonIC. When it is a CAC, it is attached to the matrix clause higher than VP and higher than Negation, but lower than Epistemic modality; hence, it is attached higher than an ottogo čto-clause. This accounts for the facts observed so far:
contrary to an ottogo čto-clause, a potomu čto-clause may not occur inside the scope of matrix negation;
it can be used both epistemically and on the speech act level, but the latter is possible only when a potomu čto-clause co-occurs with an imperative main clause or refers to a question located in a separate sentence;
it may occur inside the scope of epistemic modals;
it does not allow the relative tense marking of its predicate.
Points (i), (iii) and (iv) follow from (58) and my assumptions straightforwardly. As for (ii), it is again based on Frey’s idea of the mapping between the external syntax of a causal clause and its semantic type, according to which only NonICs can refer to a causal relation between speech acts. Since, as I assumed, a potomu čto-clause needs an indicator in order to function as a NonIC, it can be used on the speech act level only if combined with imperatives or when referring to “separated” questions, with both being indicators of the speech act use.
Finally, poskol’ku- and tak kak-clauses сan be both CACs, PACs and NonICs. When they are CACs, they are attached roughly at the same level as potomu čto-clauses, i.e., higher than negation and lower than Epistemic modality, and higher than ottogo čto-clauses. This analysis, if correct, accounts for the following facts:
neither poskol’ku- nor tak kak-clauses may occur inside the scope of matrix negation;
both can be used epistemically and as speech act clauses without any indicators being present (contrary to potomu čto);
both may occur inside the scope of epistemic modals;
neither poskol’ku nor tak kak-clause allow the relative tense marking of the predicate.
All the points in (i)–(iv) follow from (58) and my previous assumptions.
The conclusion that a poskol’ku-clause and a tak kak-clause may be as tightly integrated as a potomu čto-clause, i.e., that all three clauses may be CACs, may seem surprising at first sight given that poskol’ku and especially tak kak, contrary to potomu čto, only marginally allow focusing (for which reason Belyaev (2015b), as discussed in Section 3.2.1, deems tak kak to be semantically coordinating). However, a number of additional facts support the assumption that both poskol’ku and tak kak can indeed be CACs:
both poskol’ku and tak kak can, albeit marginally, introduce an answer to a why-question (Section 3.2.1);
both can, even though marginally, be used within the èto-focus construction (Section 3.2.1).
It can be argued, indeed, that both the capacity to serve as an answer to a why-question and to be used within the èto-focus are indices of a CAC structure. The former assumption is corroborated by the fact that a string of a why-question and an answer to it obligatorily requires an eventuality-related interpretation, exactly as CACs do according to Frey (2016). Consider the examples in (59). Both in (59a) and (59b), an answer to a why-question is introduced by potomu čto, but only the answer in (59a) is felicitous. The difference between (59a) and (59b) is that in (59a) potomu čto is used as an eventuality-related marker, while in (59b) it is used epistemically.
The second assumption, namely that being used within the éto-focus is a symptom of being a CAC, is similarly supported by the fact, already discussed in Section 3.2.3, that only eventuality-related causal clauses can participate in the èto-focus (cf. Example (24)).
To summarize, the syntactic analysis proposed above for ottogo čto-, potomu čto-, poskol’ku- and tak kak-clauses allows accounting for the whole range of formal and interpretative contrasts between them, discussed in Section 4.2, including some of those observed by Belyaev (2015b). I argue, thus, that these clause types represent different classes of adverbial clauses, and none of them (pace Belyaev 2015b) enters the class of coordinated clauses either syntactically or semantically.
5 Summarizing the oppositions encoded by Russian causal clauses
In the present Section, I unify the analysis of the four causal markers considered in Section 4 with the analysis of the marker ibo in an attempt to gain an integral picture of the system of Russian causal clauses. Ibo-clauses, as I suggested in Section 3.3, are coordinated, while potomu čto-, ottogo čto-, poskol’ku- and tak kak-clauses, according to the data in Section 4, are all subordinate adverbial clauses that differ in degree of their syntactic integration with the host clause. Further questions are linked to the relationship between coordinated clauses and NonICs: since both of them are supposed to be syntactic orphans, it remains unclear as to what the differences between them are and what the reasons are, if any, to treat them as opposed in terms of coordination and subordination.
In order to answer these questions, in what follows I compare the properties of ibo-clauses, on the one hand, and poskol’ku- and tak kak-clause, on the other, with the former being coordinated and the latter non-integrated in some of their uses. As shown in Section 5.1, the prediction is borne out that both ibo-, poskol’ku- and tak kak-clauses share features associated with syntactic orphans. However, Section 5.2 reveals that there are also features, not necessarily associated with syntactic orphans, according to which ibo-clauses differ from both poskol’ku- and tak kak-clauses. I attribute such differences to the fact that ibo-clauses are coordinated, while poskol’ku- and tak kak-clauses are adverbial clauses, which confirms the need for demarcation between the two classes. A holistic picture of properties with respect to which ibo-, potomu čto-, ottogo čto-, poskol’ku- and tak kak-clauses differ is presented. Finally, in Section 5.3 I tentatively tackle the question as to what the difference between coordinated and non-integrated clauses is from a more general perspective.
5.1 Shared properties of ibo-, poskol’ku- and tak kak-clauses
According to what was proposed so far, an ibo-clause is coordinated, hence, a syntactic orphan, while poskolku- and tak kak-clauses are syntactic orphans when used as NonICs. If this analysis is correct, it is to be expected that ibo-, poskol’ku- and tak kak-clauses pattern identically with respect to the properties associated with syntactic orphans. This expectation is indeed borne out.
An ibo-clause, like both poskol’ku- and tak kak-clauses, cannot occur inside the scope of matrix negation. In (60), the version (a) with ottogo čto differs from the version (b) with ibo in that in the former, the causal relation is inside the scope of matrix negation ((60a) denies that Russian science lost because Russian scientists took part in organizing national research centers), while in the latter it is outside the scope of negation ((60b) gives a cause for why Russian science did not lose).
Like poskol’ku and tak kak, ibo can be used both epistemically and at the speech act level. Consider examples (61) and (62) respectively. When used at the speech act level, ibo is not limited to imperative structures (cf. ibo within a question in (62)), so one may conclude that it does not require any special indicators to support the speech act reading. In this respect, again, it patterns with poskol’ku and tak kak.
An ibo-clause is also similar to poskol’ku- or tak kak-clauses in that it does not support the relative tense marking of its predicate. In (63), ibo cannot be used since the present tense of the embedded predicate možet marks simultaneity with a time point in the past encoded by the main predicate zaperežival.
Unexpectedly enough, an ibo-clause patterns differently than both poskol’ku- and tak kak-clauses as far as the scope of modals is concerned – it cannot occur inside the scope of an epistemic modal. Consider the examples in (64). (64a) with poskol’ku can express an assumption that he visited Russia for the reason that his daughter lived there, with the causal relation being inside the scope of the epistemic modal verojatno. This reading is unnatural in (64b), where ibo is used. The most feasible interpretation of (64b) is one in which the causal relation is outside the scope of the epistemic modal, i.e., (64b) gives the reason for the assumption that he visited Russia.
However, this contrast in fact does not contradict the idea that ibo-, poskol’ku and tak kak-clauses may share properties of syntactic orphans, since being inside the scope of epistemic modals is a property associated with CACs, and not with syntactic orphans. Both poskol’ku- and tak kak-clauses can be CACs, according to my assumption. In this case, they are attached to the matrix clause lower than Epistemic modality, hence, may occur inside the scope of epistemic modals (Section 4.3). An ibo-clause, on the contrary, is supposed to be coordinated, hence a syntactic orphan, in all of its uses.
The presented data are summarized in Table 6. We can see that ibo patterns identically to poskol’ku and tak kak according to all parameters except for the scope of modals. Thus, the hypothesis that ibo-clauses share with poskol’ku- and tak kak-clauses traits, associated with the absence of syntactic integration, has been confirmed.
|Cause < NEG||Cause < mod||Rel. tense||Epist. use (cause > mod)||Speech act use (without indicators)|
5.2 Contrasting properties of ibo-, poskol’ku- and tak kak-clauses and the system of Russian causal clauses in general
If it is true that ibo is a coordinating marker, while both poskol’ku and tak kak introduce adverbial clauses, there must be a formal manifestation of this distinction. I suggest that the distinction is manifested in that poskol’ku- and tak kak-clauses can precede their host clause (as in Examples (55) and (56)), while an ibo-clause must follow its host clause (cf. (27)). Importantly, both poskol’ku- and tak kak-clauses can be preposed even when they are used at the speech act level, i.e., when they are guaranteed to be NonICs according to my assumptions. In Examples (55) and (56), indeed, the preposed poskol’ku and tak kak are used as speech act conjunctions.
More generally, coordinated clauses can be assumed to differ from non-integrated adverbial clauses in that only the latter allow preposing (see Haegeman (2012: 167–168) for a similar proposal). The formal contrasts displayed by Russian causal clauses, and the oppositions that lie behind these contrasts, can thus be summarized as in Table 7.
|CACs vs. PACs vs. NonICs||Coor. vs. subor.|
|Cause < neg||Cause < mod||Rel. tense||Epist. use (cause > mod)||Speech act use (without indicators)||Preposing|
|ottogo čto (CAC)||+||+||+||−||−||+|
|potomu čto (CAC, PAC or NonIC if supported by indicators)||−||+||−||+||−||+ a|
|poskol’ku (CAC, PAC or NonIC)||−||+||−||+||+||+|
|tak kak (CAC, PAC or NonIC)||−||+||−||+||+||+|
aAlthough, as shown in Section 3.2.2, preposing is generally problematic for potomu čto, I mark potomu čto in Table 7 as compatible with preposing since, first, potomu čto can be preposed if contrastively focused, and, second, it is not the coordination-subordination distinction, but the information structure that lies behind the linear order constraint according to my assumptions.
5.3 Coordinated versus non-integrated adverbial clauses: attempt of demarcation
As both coordinated and NonICs are syntactically not integrated, it is to be clarified as to what level of language the differences between them relate to. I suggest that it is the level of information packaging in discourse. Adapting the approach to coordination according to which coordinating conjunctions introduce rhetorical relations (see Section 3.1), one may speculate that subordinating conjunctions do not introduce rhetorical relations even if they are part of a NonIC. While coordinated clauses always form independent speech acts, a NonIC may constitute a speech act not on its own, but together with its host clause. In (65) (=(56)), the non-integrated tak kak-clause does not comprise part of the interrogative speech act of the main clause, nor does it seem to constitute a declarative speech act.
It follows from this reasoning that while a coordinated clause independently participates in the rhetorical structure, a NonIC enters this structure together with its licensing clause. Compare the assumption, spelled out in (Frey 2012: 426), that a combination of a NonIC and its associate clause may describe “one single complex event”, whereas two independent clauses describe “a sequence of two events”. This may be the reason why the linear order of clauses is looser for NonICs than for coordinated clauses: the former do not interact with the discourse context on their own, hence the position they have with respect to their host clause does not affect the rhetorical relation with this context. I leave for the future a detailed elaboration of this tentative analysis.
If my proposal is on the right track, coordinated clauses can be assumed to be more loosely integrated with their host clause than NonICs. This results in an “integration hierarchy” (66), which ranks different clause types according to the degree of their integration with the host clause (from the less integrated on the left to the most integrated on the right). Note that “integration” refers to syntactic structure with respect to three upper nodes of the hierarchy (66) and to discourse structure with respect to two lower nodes.
In the literature on causal clauses, attempts have been made to account for the syntactic and interpretative contrasts between different clause types in terms of coordination and subordination. My analysis of five Russian causal markers (potomu čto, ottogo čto, poskol’ku, tak kak and ibo) suggests that only one marker (ibo) is coordinating. Other markers introduce adverbial (hence, subordinate) clauses, while the contrasts between them are due to different degrees of syntactic integration with the host clause, associated with different classes of adverbial clauses (CACs, PACs and NonICs). An ottogo čto-clause can only be a CAC; a potomu čto-clause is by default a CAC or a PAC, but if it is supported by indicators it can also be a NonIC; poskol’ku- and tak kak-clauses are the least integrated ones – they can freely be CACs, PACs and NonICs. A consistent picture of Russian causal clauses thus cannot be obtained unless both the coordination-subordination distinction and the distinction between classes of adverbial clauses are taken into consideration.
The phenomena that turned out to be useful when comparing causal CACs, PACs and NonICs in Russian regard scope of matrix negation, scope of epistemic modals, the possibility of the relative tense marking and the epistemic and speech act use of conjunctions. Some of these phenomena were treated in the literature – erroneously, in my view – as criteria of coordination and subordination. On the contrary, the possibility of positioning a causal clause before the host clause, which is another well-known criterion of coordination and subordination, turned out to be helpful indeed for differentiating adverbial clauses and coordinate clauses.
While the contrast between CACs, PACs and NonICs seems to be typologically valid as far as the systems of causal clauses are concerned, the realizations of this contrast are language specific. Russian possesses an opposition of two causal CACs, similar both semantically and morphosyntactically, one of which (an ottogo čto-clause) is base generated in a deeper position than the other (a potomu čto-clause). I am not aware of this kind of opposition in English, German or French, languages that have been said to manifest formal contrasts between causal clauses similar to those that one finds in Russian. Furthermore, different degrees of syntactic integration underlie the formal differences between the German weil and da, on the one hand, and the Russian potomu čto and poskol’ku, on the other, although, at first sight, these two pairs seem to be very close counterparts. Both the Russian potomu čto- and poskol’ku-clauses can be CACs, while the German da-clause can only be a PAC or a NonIC according to Frey (2016).
The analysis I have proposed also raises a more general question as to what level of language the difference between coordinated and non-integrated adverbial clauses relates to, with both being syntactic orphans. I suggest tentatively that coordinated clauses are less integrated with the licensing clause than NonICs in terms of discourse structure. A kind of integration hierarchy thus emerges: coordinated clauses < NonICs < PACs < CACs.
- PP –
Funding source: Ministry of Science and Higher Education of the Russian Federation
Award Identifier / Grant number: 075-15-2020-793
I am profoundly thankful to the participants of the seminars on corpus-based analysis of Russian grammar, held in 2011–2019 in the V.V. Vinogradov Russian Language Institute of the Russian Academy of Sciences, for their insightful questions and fruitful discussion. I would also like to thank the anonymous Linguistics reviewers, who provided numerous helpful comments and corrections. All errors remain mine.
Research funding: The work was supported by the Ministry of Science and Higher Education of the Russian Federation within Agreement No 075-15-2020-793.
Data source: Russian National Corpus (RNC) www.ruscorpora.ru.
Belyaev, Oleg. 2015a. Systematic mismatches: Coordination and subordination at three levels of grammar. Journal of Linguistics 51. 267–326. https://doi.org/10.1017/s0022226714000450.Search in Google Scholar
Belyaev, Oleg. 2015b. Cause in Russian and the formal typology of coordination and subordination. In Peter Arkadiev, Ivan Kapitonov, Yury Lander, Ekaterina Rakhilina & Sergey Tatevosov (eds.), Donum semanticum: Opera linguistica et logica in honorem Barbarae Partee a discipulis amicisque Rossicis oblata, 35–66. Moscow: Jazyki slavjanskix kul’tur.Search in Google Scholar
Boguslavskaja, Olga Ju & Irina B. Levontina. 2004. Smysly ‘pričina’ i ‘cel’’ v estestvennom jazyke [Meanings ‘reason’ and ‘purpose’ in natural language]. Voprosy jazykoznanija 2. 68–88.Search in Google Scholar
Brext, Ričard D. 1985. O vzaimosvjazi meždu nakloneniem i vremenem: sintaksis časticy by v russkom jazyke [On the relationship between mood and time: syntax of the particle by in Russian]. In Tatjana V. Bulygina & Aleksandr E. Kibrik (eds.), Novoe v zarubežnoj lingvistike. Vypusk XV. Sovremennaja zarubežnaja rusistika [New Research in Foreign Linguistics. vol. XV. Contemporary Russistics Abroad], 101–117. Moscow: Progress.Search in Google Scholar
Bulachovsky, Leonid A. 1954. Russkij literaturnyj jazyk pervoj poloviny XIX veka. Fonetika. Morfologija. Udarenie. Sintaksis. [Russian literary language of the first half of the XIX century. Phonetics. Morphology. Stress. Syntax.]. Мoscow: Učpedgiz.Search in Google Scholar
Cinque, Guglielmo. 1999. Adverbs and functional heads: A cross-linguistic perspective. Oxford: Oxford University Press.Search in Google Scholar
Comrie, Bernard. 2008. Subordination, coordination: Form, semantics, pragmatics. In Edward J. Vajda (ed.), Subordination and coordination strategies in North Asian languages (Current Issues in Linguistic Theory 300), 1–16. Amsterdam & Philadelphia: John Benjamins.10.1075/cilt.300.03comSearch in Google Scholar
Cristofaro, Sonia. 2003. Subordination. Oxford: Oxford University Press.Search in Google Scholar
Endo, Yoshio. 2012. The syntax-discourse interface in adverbial clauses. In Lobke Aelbrecht, Liliane Haegeman & Rachel Nye (eds.), Main clause phenomena: New horizons, 365–383. Amsterdam & Philadelphia: John Benjamins.10.1075/la.190.16endSearch in Google Scholar
Evgen’eva, Anastasija P. (ed.). 1985. Slovar’ russkogo jazyka (Malyj akademičeskij slovar’) [Dictionary of Russian (Small Academic Dictionary)], vol. 1. Moscow: Russkij jazyk.Search in Google Scholar
Fasmer, Maks. 1986. Ètimologičeskij slovar’ russkogo jazyka. [Etymological dictionary of Russian], vol. 2, 2nd edn. [Muza A. Oborina (ed.)]. Moscow: Progress.Search in Google Scholar
Frey, Werner. 2012. On two types of adverbial clauses allowing root-phenomena. In Lobke Aelbrecht, Liliane Haegeman & Rachel Nye (eds.), Main clause phenomena: New horizons, 405–429. Amsterdam & Philadelphia: John Benjamins.10.1075/la.190.18freSearch in Google Scholar
Frey, Werner. 2016. On some correlations between formal and interpretative properties of causal clauses. In Ingo Reich & Augustin Speyer (eds.), Co- and subordination in German and other languages (Linguistische Berichte Sonderhefte 21), 153–179. Hamburg: Helmut Buske Verlag.Search in Google Scholar
Groupe λ-l. 1975. Car, parce que, puisque. Revue Romane 10(2). 248–280.Search in Google Scholar
Haegeman, Liliane. 2012. Adverbial clauses, main clause phenomena and the composition of the left periphery: The cartography of syntactic structures (Oxford Studies in Comparative Syntax 8). Oxford: Oxford University Press.10.1093/acprof:oso/9780199858774.001.0001Search in Google Scholar
Haspelmath, Martin. 1995. The converb as a cross-linguistically valid category. In Martin Haspelmath & Ekkehard König (eds.), Converbs in cross-linguistic perspective (Empirical Approaches to Language Typology 13), 1–55. Berlin & New York: Mouton de Gruyter.10.1515/9783110884463-003Search in Google Scholar
Iordanskaja, Lidija N. 1988. Semantika russkigi sojuza raz (v sravnenii s nekotorymi drugimi russkimi sojuzami) [Semantics of the Russian conjunction raz (in comparison to some other Russian conjunctions)]. Russian Linguistics 12(3). 239–267.Search in Google Scholar
Kazenin, Konstantin I. & Yakov G. Testelets. 2004. Where coordination meets subordination. In Martin Haspelmath (ed.), Coordinating constructions, 227–241. Amsterdam & Philadelphia: John Benjamins.10.1075/tsl.58.13kazSearch in Google Scholar
Kobozeva, Irina. 2000. The problem of identification and syntactic representation of Russian complex sentences with illocutionary-independent subordinate clauses. In Gerhild Zybatow, Uwe Junghanns, Grit Mehlhorn & Luka Szucsich (eds.), Linguistische Arbeitsberichte 75 (3. Europäische Konferenz “Formale Beschreibung slavischer Sprachen”, Leipzig 1999), 67–79. Leipzig: Institut für Linguistik/Universität Leipzig.Search in Google Scholar
Krifka, Manfred. 2013. Response particles as propositional anaphors. In Snider Todd (ed.), Semantics and linguistic theory (SALT) 23, 1–18. Washington, D.C.: LSA Open Journal Systems.10.3765/salt.v23i0.2676Search in Google Scholar
Latysheva, Alla N. 1982. O semantike uslovnyh, pričinnyh i ustupitel’nyh sojuzov v russkom jazyke [On semantics of conditional, causal and concessive conjunctions in Russian]. Vestnik Moskovskogo Universiteta 5(serija Filologija). 51–59.Search in Google Scholar
Lehmann, Christian. 1988. Towards a typology of clause linkage. In John Haiman & Sandra A. Thompson (eds.), Clause combining in grammar and discourse (Typological Studies in Language 18), 181–226. Amsterdam & Philadelphia: John Benjamins.10.1075/tsl.18.09lehSearch in Google Scholar
Paducheva, Elena V. 1977. Ponjatie prezumpcii v lingvističeskoj semantike [The notion of presupposition in linguistic semantics]. Semiotika i informatica 8. 91–124.Search in Google Scholar
Padučeva, Elena V. 1990. Meždu predloženiem i vyskazyvaniem: sub’ektivnaja modal’nost’ i sintaksičeskaja nepodčinimost’ [Between sentence and utterance: subjective modality and root phenomena]. Revue des études slaves 62(1–2). 303–320. https://doi.org/10.3406/slave.1990.5887.Search in Google Scholar
Pekelis, Olga E. 2009. Sočinenie i podčinenie v kontekste pričinnoj semantiki [Coordination and subordination in the context of causal semantics]. Moscow: Russian State University for the Humanities Cand. thesis.Search in Google Scholar
Pekelis, Olga E. 2015. Coordination and subordination [Sochinenije i podchineije]. Materialy dlja proekta korpusnogo opisanija russkoj grammatiki [Towards a corpus description of Russian grammar]. Moscow: Manuscript. http://rusgram.ru (accessed 29 May 2020).Search in Google Scholar
Pekelis, Olga E. 2017. Pričinnye pridatočnye [Causal clauses]. In Vladimir Plungjan & Natalya Stojnova (eds.), Materialy k Korpusnoj grammatike russkogo jazyka: sintaksičeskie konstrukcii i grammatičeskie kategorii [Towards a corpus-based Russian grammar: syntactic constructions and grammatical categories], vypusk II, 55–131. Saint Petersburg: Nestor-Istorija.Search in Google Scholar
Rizzi, Luigi. 1997. The fine structure of the left periphery. In Liliane Haegeman (ed.), Elements of grammar: A handbook of generative syntax, 281–337. Dordrecht: Kluwer.10.1007/978-94-011-5420-8_7Search in Google Scholar
Sohmiya, Yoshikazu. 1975. Syntax and Semantik der kausalen Konjuktionen, denn and weil. Doitsu Bungaku 42. 108–119.Search in Google Scholar
Šnittke, Ekaterina L. 2020. K voprosu o soglasovanii vremen v sovremennom russkom jazyke: korpusnoe issledovanie distributivnyx xarakteristik vremennyx form v sentencialnyx aktantax [Towards the problem of sequence of tenses in modern Russian: A corpus study of distributional characteristics of tense forms in complement clauses. Voprosy jazykoznanija 3. 26–51.Search in Google Scholar
Švedova, Natalja Yu (ed.). 1980. Russkaja grammatika [Russian Grammar], vol. II. Moscow: Nauka.Search in Google Scholar
Uryson, Elena V. 2011. Opyt opisanija semantiki sojuzov: lingvisticheskie dannye o dejatel’nosti soznanija [An attempt at describing the conjunctions’ semantics: linguistic data of consciousness activity]. Moscow: Jazyki Slavjanskih Kul’tur.Search in Google Scholar
Weiss, Daniel. 1989. Parataxe und Hypotaxe: Versuch einer Skalarisierung. In Wolfgang Girke (ed.), Slavistische Linguistik 1988, 287–322. Munich: Sagner.Search in Google Scholar
Zaika, Natalia M. 2019. Polipredikativnye pričinnye konstrukcii v jazykah mira: prostranstvo tipologičeskih vozmožnostej [Polypredicative reason constructions in the world’s languages: Typological parameters]. Voprosy jazykoznanija 4. 7–32. https://doi.org/10.31857/s0373658x0005702-1.Search in Google Scholar
Zaliznjak, Anna А. 1992. Issledovanija po semantike predikatov vnutrennego sostojanija. [Studies on the semantics of predicates of the internal state]. Munich: Sagner.Search in Google Scholar
Zemskaja, Elena A. (ed.). 1978. Russkaja razgovornaja reč [Russian colloquial speech]. Moscow: Nauka.Search in Google Scholar
© 2022 the author(s), published by De Gruyter, Berlin/Boston
This work is licensed under the Creative Commons Attribution 4.0 International License.