I. M. Kobozeva (Moscow State University Lomonosov)

I. M. Kobozeva (Moscow State University Lomonosov)

The problem of identification and syntactic representation of Russian complex sentences with illocutionary – independent subordinate clause

In this paper I analyze Russian complex sentences (CSs), distinguished by their functioning as a form of complex speech acts, that is speech acts consisting of two or more elementary speech acts. We call CSs of this kind pragmatically complex and characterize their subordinate clauses as illocutionary – independent, because they correspond to a separate speech act with its own illocutionary force (IF), that is different from IF of a main clause.

In Russian grammatical tradition CSs were not considered from pragmasemantic point of view and that is why the phenomenon of pragmatically complex CS has not been recognized, not to say analyzed. So in Section 1 of the paper I demonstrate the existence of CSs with illocutionary – independent subordinate clauses, in Section 2 I discuss semantic and syntactic criteria of illocutionary – independence of clauses in CS and draw conclusions for syntactic representation of pragmatically complex CSs and subcategorization of subordinating conjunctions.

1. Can a subordinate clause have its own IF?

Before we answer this question we must first of all define more exactly the notion of CS, accepted in this paper, and secondly explain our understanding of relationship between syntactic structure of a sentence and its IF.

Here I accept traditional understanding of CS as a sentence having another sentence as its constituent, the one that is “attached” to some other constituent by some subordinating conjunction or “conjunction word”. (In generative grammar traditional conjunctions are categorized either as complementizers, as e.g. èto ‘that’, or prepositions, as e.g. poka ‘while’ and “conjunction words” are treated as wh-words moved to the position of COMP). The syntactic structure of CS in terms of trees of immediate constituents is roughly represented in (1):

(1)S

...CP/PP...

COMP/PS

Subordinate clause is a sentential constituent of CS that contains a lexically represented subordinating device. It corresponds to CP/PP in (1).

As for relations between syntactic structure of a sentence and IF of an utterance produced by using this sentence, I follow performative hypothesis only to the extent that I also believe, that the predestination of a sentence to be used with a certain range of IFs (i.e. its illocutionary potential) is represented in its syntactic structure by a special element corresponding to the grammatical notion «type of a sentence according to the purpose of the utterance» in traditional rusistics and to «syntactic type of a sentence» in western grammatical terminology. In generative paradigm «type according to purpose» can be treated as a separate functional category of illocutionarity (Illoc). Its projection IllocP corresponds to a sentence with a fixed illocutionary potential and is a complement to COMP. Illoc contains grammatical illocutionary features and its complement is so to say a «forceless» sentence, i. e. an inflectional phrase – IP. For the purpose of the given study the number and specific character of grammatical illocutionary features is not important (as possible candidates one could envisage four elementary features: narrativity (Nar), interrogativity (Q), imperativity (Imp) and exclamation (Excl)). Illocutionary features interrelate with mood features, trigger movement and determine the choice of intonation pattern. Performative verbs (such as state, ask, promise, etc.) are lexically marked as [+ Illoc] and at the level of logical form they move to the position of Illoc in the appropriate context of 1-st person and present tense. Conceptual-intentional system interprets illocutionary features, combining their meaning with meanings of other modal elements of a sentence (e. g. modal particles, cf. [Kobozeva 1989]) and with lexical meanings of performative verbs thus forming an illocutionary part of a semantic representation of a sentence. In order to account for the actual IF of an utterance, produced by a given sentence in a given situation, we need to turn to pragmatics, e.g. to speech act theory. According to this treatment of relationship between IF and syntactic structure the question opening this section can be reformulated as follows: does any sentence irrespective of its syntactic complexity have only one instance of Illoc or Illoc can iterate in sentential structure, appearing within some lower CP; in other words, are structures like (2) possible:

(2)CP

COMPIllocP

…ÑP…

COMPIllocP

As neither traditional nor modern grammar provide an answer to this question, we have to turn to linguistic data and study illocutionary characteristics of clauses within CSs.

There is a prejudice, that an utterance, expressed by CS, has only one IF and in this respect it is not different from an utterance, expressed by a simple sentence. It is true, that many types and kinds of CSs correspond to elementary speech acts, i.e. to speech acts, having a single IF. Among them CSs with complement clauses, CSs with restrictive relative clauses, different kinds of CSs with circumstantial clauses (temporal, local, conditional, purpose clauses among them) exemplified in (3), (4) and (5) respectively:

(3)Vy znaete, èego javamželaju.

youknowwhat-genIyou-datwish

‘You know what I wish you.’

(4)Gde ostanovka avtobusa, kotoryjidet v Poreè’e?

wherestop-nombus-genwhichgoestoPoreè’e

‘Where is the stop of the bus, that goes to Poreè’e?’

(5) Pokažiemuètopis’mo,ètoby onznal vsju pravdu.

showhim this letter, that+subj.clitiche knewwholetruth.

‘Show him the letter in order that he knew the whole truth.’

CS (3) is intended for speech acts with IF of representative type [Searle 1976]. Although (3) contains a complement clause in interrogative form (called «indirect interrogative» in traditional Russian syntax), this clause does not correspond to a separate speech act with IF of a question, serving only as a means for nominating a referent, thus participating in the act of reference. In speech act theory act of reference is treated as an integral part of a propositional act [Searle 1965] and according to Austin it belongs to locutionary aspect of speech act (as opposed to illocutionary). CS (4) also has only one IF – that of a question – and its relative clause again serves referential purposes, functioning as a part of a definite description that restricts the domain of possible referents of the NP it modifies. CS (5) represents an elementary speech act with IF of a directive type expressed in the main clause. As for the subordinate purpose clause, it does not constitute a separate speech act, it just modifies the propositional content of a directive expressed by the main clause by indicating the goal of the action that the addressee is impelled to perform.

Such a situation may seem the only one possible for CS, because just like a simple sentence, in discourse it corresponds to an utterance – a unit of speech perceived as something whole. On the one hand the integrity of CS is indicated by its surface characteristics (presence of a conjunction, prosody), on the other hand postulated uniqueness of IF in CS conforms to the commonly held view that an utterance has a unique communicative purpose/intention (cf. Bakhtin 1979, Shirjajev 1999). What’s more, for some types of CSs the uniqueness of IF is determined by the structure of their propositional content. Since only a full proposition or propositional form can be “the carrier” of IF, the main clause of CSs with subject or complement subordinate clauses cannot have its own IF, separated from the IF of the subordinate clause, because it cannot constitute a full proposition, when it lacks one of its obligatory arguments. However if we take the full range of possible CS types in Russian the generalization about the uniqueness of IF does not hold. It is demonstrated by the following CSs, whose constituent clauses function as separate speech acts each with its own IF:

(6)Jaživuxorošo,èegoivamželaju.

I livewellwhat-genandyou-datwish-1p/sg

‘I am fine and wish you the same.’

(7)Zavtra priedet Ivanov,kotoromujaprošutebjaperedat’èto pis’mo.

tomorrowcomes-futIvanovwhich-datIaskyouto givethis letter

‘Ivanov will come tomorrow and I ask you to give him this letter.’

(8)Emu uželuèše,takètostoit li vyzyvat’vraèa?

hå-dat alreadybetter,sothatworthwhetherto call fordoctor-acc

‘He feels better now, so is it necessary to call for a doctor?’

According to traditional formal criteria sentences (6) – (8) must be considered as CSs, but in all of them subordinate clauses correspond to a separate speech act, having IF different from IF of the main clause. The main clause in CSs (6), (7) è (8) have an IF of a representative type, while the subordinate clause in (6) has IF of a wish (expressive type), the relative clause in (7) has a directive IF of a request and the subordinate clause of entailment in (8) has an IF of a question casting doubt on the appropriateness of some putative action. It is true, that illocutionary independence of speech acts, corresponding to constituent clauses of CSs like (6) – (8) is not absolute. These speech acts are tied together closely enough to be expressed by an integrated although complex syntactic structure. It is just the existence of pragmatic (illocutionary) relations, analogous to dependency relations in syntax (see e. g. the notion of illocutionary necessitation [Baranov, Krejdlin 1992]), that served as a ground for the introduction of the concept complex speech act in linguistic pragmatics.

What enables us to assign to subordinate clauses in CSs of the type, exemplified by (6) – (8) their own IFs and thus consider them as illocutionary independent representatives of separate elementary speech acts, although integrated into complex speech acts? In (6) and (7) the illocutionary independence of subordinate clauses is made evident by the presence of performatively used illocutionary verbs (želaju‘wish’, prošu ‘ask’), that explicate their IF, that is different from main clause IF. The fact that these verbs are used performatively is essential. Indeed, uttering (6) or (7) in appropriate circumstances the speaker really performs speech acts of wishing and requesting respectively and not just informing. Illocutionary independence of the subordinate clause in (8) from its main clause is indicated by the presence in the former of overt illocutionary markers of interrogativity, namely interrogative particle li «whether» together with the special intonation contour (IC-3 in E. A. Bryzgunova’s classification [RG 1980]), in the written form reflected by the question mark «?», while the latter is narrative. Full grammaticality of examples like (6) – (8) shows that syntactic structures of the type (2) with illocutionary independent embedded clauses are possible (at least for Russian). Now the problem arises: how to distinguish “forceless” embedded clauses from illocutionary independent ones in case when the clause has no overt illocutionary markers? This is the case of non-performative narrative subordinate clauses, for «narrativity» as a feature of category Illoc has no overt markers[1] in Russian as in many other languages. So independent sentences with Illoc feature Nar superficially are no different from the corresponding embedded clauses that do not have Illoc in their structure (if we neglect conjunction, of course).

Consequently the presence in a subordinate clause of overt markers of IF – either lexical (illocutionary verbs in performative form) or grammatical (interrogative or imperative) – is a sufficient but not necessary condition of its illocutionary independence. Are there any ways to decide whether a given subordinate clause is illocutionary independent in cases when there are no overt illocutionary markers in its lexico-syntactic structure? I propose two criteria of illocutionary independence of a subordinate clause without overt illocutionary markers, that can be used for identification of pragmatically complex CSs.

2. Criteria of illocutionary independence of a subordinate clause

2.1. Semantic criterion: semantic structure of an illocutionary independent subordinate clause must contain an assertive component.

Propositional content of any utterance contains an assertive component (an assertion or its analogue: what is asserted, what is asked about or asked for etc.). Consequently, a CS can be pragmatically complex only if each of its two parts, i.e. the main and the subordinate clause, has an assertive component in its semantic structure. In other words the content of an illocutionary independent subordinate clause cannot be presupposed or non-asserted (alleged) as a whole[2]. For formally «narrative» CSs the presence of assertion in a clause is verified by a negation test, that is generally used to discover presuppositios and other non-asserted components of meaning. If we put CS under general negation and find out that it unambiguously denies only one proposition and this proposition is not the one expressed by the subordinate clause, it means that this clause does not contain any assertion and as such it cannot be illocutionary independent. We get such a result in two cases: 1) when the value «true» of the subordinate proposition remains unchanged under negation and thus it constitutes a presupposition of the tested CS; 2) when the truth value of the subordinate proposition in the negated CS is suspended and thus it constitutes an allegation of tested CS. If CS is ambiguous between readings with different scope of negation, it means that both main and subordinate clauses have assertive components in their semantic structure and so the latter is illocutionary independent. Compare (9) – split CS with causal clause (according to Latyševa1989) and (10) – its non-split counterpart:

(9)Onostalsjadomapotomu,ètoonzabolel.

hestayedhomefor the reasonthathefell ill

‘The reason for his staying home was that he fell ill.’

(10)Onidoma,potomu ètooknaotkryty.

theyhomefor the reasonthatwindowsopened

‘They are at home, because the windows are open.’

Negating (9), we do not deny propositions of the main clause ‘he stayed home’ or of the subordinate clause ‘he fell ill’, because these two propositions are not asserted in (9): the proposition of the main clause ‘he stayed home’ is a presupposition of (9) and thereby escapes negation; the proposition of the subordinate clause ‘he fell ill’ is an allegation of (9), under negation it is not denied, its truth value is just suspended. What is denied, is the one and only assertion of (9): the reason for his staying home was that he fell ill. Having only one assertion, CS (9) can have only one IF and so it is pragmatically simple. The negation of (10) is ambiguous: it can deny any of the three propositions: 1)that they are at home; 2) that the windows are open; 3) that the truth of 2 can be taken as a ground for believing in the truth of 1. The source for such an ambiguity of the scope of negation is that the negated sentence includes more than one assertion (see e. g. Paduèeva 1974: 156-157) and consequently it may be a pragmatically complex CS. Applying general negation test to CSs (3) and (4) that obviously have only one IF in their semantic structure we get the expected unambiguous negation, that correctly predicts that these CSs cannot be pragmatically complex. Unfortunately, negation test applies only to CSs with “narrative” main sentences, so pragmatic complexity of CSs like (8) cannot be checked by this test. In such cases we can use another criterion, that does not constraint illocutionary type of the main clause – a syntactic one.

2.2. Syntactic criterion: illocutionary independent subordinate clause must admit at least some of “main clause phenomena”.

“Main clause phenomena” is a term that was introduced for linguistic units of different nature (word, idiom, syntactic construction, grammatical marker or intonation pattern) that seemed to occur only in independent simple sentences or main clauses of CSs [Hooper, Thompson 1973; Green, 1976; Bolinger 1977]. For Russian such units were singled out in [Paduèeva 1990], where their characteristic syntactic feature was called “syntactic unsubordinability”. In a later paper [Paduèeva 1996: 299] an important qualification was made concerning syntactic unsubordinability of “main clause phenomena” (further referred to as MCP). E.V.Paduèeva correctly observed that some MCP occur in certain kinds of subordinate clauses, e.g. in clauses introduced by postpositive poskol’ku ‘since’, potomu èto ‘because’, xotja ‘although’ and appositive kotoryj ‘which’ and she advanced a plausible hypothesis that such subordinate clauses correspond to separate speech acts.

This hypothesis is corroborated by the fact, that all of the mentioned kinds of subordinate clauses admit some of overt illocutionary markers, see e.g. appositive relative clause in (7), that contains performatively used illocutionary verb, indicating IF of a separate speech act, corresponding to this clause. That the verb prošu ‘ask’ is used performatively in (7) (i.e. like illocutionary marker) can be demonstrated by changing its form from indicative imperfective to subjunctive perfective poprosil by ‘would have asked’. Such a change does not alter the truth conditions of (7), it just affects its pragmatics, making (7) more polite, and this is possible only in case of performatively used verb prosit’ (and some other verbs of the same illocutionary type, e.g. sovetovat’ ‘advise’, predlagat’ ‘propose’). If the same verb form was used in complement clauses like (11):

(11)Onznaet,ètojaprošuvaspomoè’mne.

he knows that Iask-1p/sgyou-acc to help me.

‘He knows that I ask you to help me.’

or in restrictive relatives like (12):

(12)Vypoterjaliimennotuknigu,kotorujujaprošuvas èitat’ každyj den’.

youlostjustthatbook,whichIaskyou to read every day

‘You lost just that very book, which I ask you to read every day.’

the same change would affect the truth conditions of the sentence, cf.:

(11)Onznaet,ètojapoprosil byvaspomoè’mne.

he knows that Iask-subj/ms/sgyou-acc to help me.

‘He knows that I would have asked you to help me’

(12)Vypoterjaliimennotuknigu, kotoruju ja poprosil by vas èitat’ každyjden’.

youlostjustthatbook,whichIask-subj/ms/sgyou to read every day

‘You’ve lost just that very book, that I would have asked you to read every day.’

If the world is such that in the period of time including the time of the utterance the speaker did not ask the addressee to help, (11) would be necessarily vacuous, while (11) could be true. Analogously, in the world, where in the period of time including the time of the utterance the speaker did not ask the addressee to read any book every day, (12) would be necessarily vacuous, while (12) could be true. Such a change of meaning shows that in (11) and (12) the illocutionary verb is used descriptively, or constatively, as it should be, since the subordinate clauses in which they feature, do not possess IF of their own, whose overt marker this verb could be, if they did.

According to the treatment of relationship between IF and syntactic structure of a sentence outlined in section 1, we could reformulate the hypothesis about the connection between the occurrence of MCP in subordinate clauses and the illocutionary independence of such a clause as follows: MCP is licensed by an Illoc category that is not divided from it by CP bounding node.

Y. Vasil’ev [Vasil’ev 1999] tested 20 kinds of Russian MCP, listed in [Paduèeva 1990], as to their ability to occur in different types of subordinate clauses. His results confirm the hypothesis.

As was predicted by the hypothesis, no MCP is possible within just those types of subordinate clauses, that cannot have IF of their own in principle, because at the level of semantic representation they are mapped onto integral parts of the proposition, embodied by the main clause. Such are complement, temporal and restrictive relative clauses. Cf. CSs in (13), that are ungrammatical just because there is some MCP (marked by bald type throughout) in their subordinate clauses. That this is not just semantic anomaly is demonstrated by the fact, that substitution of MCP for a semantically equivalent syntactically subordinable expression (underlined) gives a semantically acceptable sentence, as corresponding CSs in (14) show: