Časopis Slovo a slovesnost
en cz

Basic problems of the integral description of language

Ruselina Nicolova



Základní problémy integrálního popisu jazyka

Contemporary linguistic theories differ considerably in their presentation of the interrelations between vocabulary and grammar as well in the space they devote to vocabulary in theory. In this respect there are two poles of theory – one of which is the theories in which vocabulary plays hardly any role at all or is of minor importance (e.g. Relational Grammar or Generalized Phrase Structure Grammar). The other pole is the theories in which certain rules of syntactic character find a place in the lexis (e.g. Lexical-Functional Grammar and Head-Driven Phrase Structure Grammar). Somewhere between these two theories lie theories such as the Government and Binding Theory and Role and Reference Grammar (RRG) that apply lexical and syntactic rules and principles (for a more comprehensive discussion see Van Valin – Wilkins, 1993, p. 501–502).

Of particular importance in Slavic studies is the ‘Meaning-Text’ model advanced by Mel’čuk and further developed by an entire group of linguists, and by Apresjan, whose fundamental works are of great weight, too. Apresjan states that ‘the vocabulary is an inseparable part of the complete theoretical description of language and is in essence just as important as grammar’ (Apresjan, 1995, p. 10). The Polish School of Semantics, A. Wierzbicka (1972, 1980), A. Bogusławski (1988, 1997) and others, is prominent. Wierzbicka has developed an exhaustive programme for the integral description of vocabulary and grammar by means of semantic primitives from the natural language. Other significant works are those of Fr. Daneš (1968) and Daneš, Hlavsa, Kořenský (1973) on Czech language, St. Karolak (1984) on Polish, D. Staniševa (1985) on Bulgarian and other Slavic languages and M. Koritkowska (1992) on a comparative level of Bulgarian and Polish (this work includes a detailed synopsis of Slavic studies literature on the topic).

With the development of linguistics in mind, I must mention that, in the next century, the concept of integral description of language in which vocabulary is given the same space as grammar, is likely to become determinant since it is gaining impetus and support among representatives of various linguistic schools. This is why the question of which methodology is most suitable for the integral description of language is of great concern. The comparison of different methods is undeniably of much interest to theory. So in this article we will review the RRG methodology by comparing it with that of the Moscow School of Semantics. We will do this by introducing a pragmatic component in the analysis of Russian and Bulgarian sentences containing lexical items identical to the English remember, since constructions with this English verb and its equivalent in the Australian language Mparntwe Arrernte are the topic of Van Valin – Wilkins (1993) and thus provides a methodology for the analysis.

The relation between lexis and syntax has various presentations in the different theories. Some of the theories that do not make any generalizations when describing [204]lexemes, e.g. a predicate, present the syntactic or semantic-syntactic valence of the lexical item. Other theories (e.g. the Government and Binding Theory, ‘Meaning-Text’ model and RRG) aspire to make a generalization in the deduction of the syntactic valence of the lexical meaning of the word. Apresjan has stressed the opposite relation, too – syntactic determination of the meaning (Apresjan, 1995, p. 538–553). In this context the presentation of the lexical meaning is of great importance.

Some theories (e.g. Montague Grammar, Dowty excepted, Lexical-Functional Grammar, Generalized Phrase Structure Grammar) do not utilize decomposition of the lexical meaning – the predicates, for example, are represented as logical constants like kill (x,y). Other theories employ decomposition or paraphrase in the presentation of the meaning (e.g. Wierzbicka, 1969, 1972 and Apresjan, 1995). One should not forget that the Prague School was the first to propose decomposition of the morphological meaning by using differential semantic features and the school has thereby had a strong influence on the further development of grammatical and lexical semantics in an array of contemporary theories – Field Theory, Generative Semantics, Head-Driven Phrase Structure Grammar etc.). Apart from this, in his renowned works on the declension in Russian, R. Jakobson posed a highly important question: the necessity of an analysis of the syntactic contexts of the case forms in order to define the combinatory (lexically dependent) varieties of the cases (Jakobson, 1936, p. 1962–1988). Through this analysis, Jakobson reached the conclusion that there exists not a six-member, but an eight-member case system in Russian by adding two cases: a genitive II, which exists only in masculine nouns with material meanings – cvet sachara: kusok sacharu, and a locative II which is used with a locational meaning – chodit’ v lesu dumat’ o lese. It was Apresjan that introduced the concept – trivial semantic feature which is of particular theoretical value to semantic decomposition in the description of the interrelation between lexis and syntax. This concept is also used in the surface-semantic rules – grammeme-lexical and lexico-lexical. An example of a non-trivial semantic features is the phase of the verbs načinat’, puskat’sja, perestavat’, prekraščat’, prodolžat’ etc., which govern the forms (V), INF., IMPERF., but not the forms (V), INF., PERF. Compare On načal, perestal, prodolžal delat’ gimnastiku not: sdelat’ gimnastiku (Apresjan, 1995, p. 25, 28–34). Unlike Wierzbicka, Apresjan stresses that ‘when analyzing an item x, one should reduce it to a „direct semantic component“ p, and not to the smallest word-senses’ (Apresjan, 1995, p. 25).

According to Van Valin – Wilkins, there are two approaches to decomposition or paraphrase: the minimalist: for instance in Z. Vendler’s (1967) four-way distinction of English predicates and in Dowty’s (1979) modified version of this classification in which the activity and state predicates are considered primitives out of which, by using operators such as BECOME, CAUSE, NOT, the predicates’ achievements and accomplishments can be drawn. This approach was adopted in the first version of RRG. It differs from Wierzbicka’s ‘rich’ semantic approach that gives a quite detailed paraphrase of the meaning by employing a vocabulary of semantic primitives from the natural language. Here, we do not have the possibility to discuss Wierzbicka’s approach, nor the positive and negative effects of using either a special terminology [205]or the vocabulary of the natural language in describing lexical meaning. I will merely note that in the RRG we are now discussing, a middle approach to semantic decomposition has been adopted for separate lexemes, but not for large groups of verbs. The RRG approach utilizes a combined metalanguage which includes semantic primitives from the natural language as in W. Dixon (1971), as well as operators from Dowty’s metalanguage such as BECOME, CAUSE, NOT. The main feature of this metalanguage is the semantic redundancy rules that explicate the syntax of the decomposed elements, the generalizations of the arguments of the basic predicates of metalanguage and more. RRG shows how semantic representation of the predicates provides a possibility for predicting its syntactic attributes considering also the numerous independently motivated, specific to a given language and universal, semantic, lexical and morphosyntactic principles. The idea of these principles is parallel with Apresjan’s concept for a large number of different factors ‘semantic, syntactic, morphological, combinatory, stylistic and even phonetic, on which the realization of the meaning of language items depends’ (Apresjan, 1995, p. 62). RRG also proposes an algorithm for linking of semantic representation with the syntactic structures through Interclausal Relations Hierarchy. This algorithm is based on typological generalizations.

As Van Valin and Wilkins note, according to Vendler’s (1967) renowned semantic classification modified and expanded by Dowty (1979) with the grammatical features of English verb classes, the English verb remember is used both as an activity and state predicate. Slavic languages, unlike English, have separate lexical items for activity and state predicates, for example pomnja in Bulgarian, pomnit’ in Russian in the imperfective aspect – state predicate. Spomnjam si in Bulgarian and vspominat’ in Russian in the imperfective aspect – activity predicate, which usually connotes an activity not controlled by the subject. This activity can be represented in the semantic structure without the DO operator for actively controlled activity adopted by Dowty (1979) (Apresjan has also noted a correlation between the features ‘momentary’ and ‘uncontrollable’ activity, see Apresjan, 1995, p. 228–229), spomnja si in Bulgarian and vspomnit’ in Russian in the perfective aspect – achievement.

The common element in the semantic structure of the indicated verbs in Bulgarian and Russian is the same as in the three varieties of the English verb remember: .think.again (x) about something.be.in.mind.from.before (y). In this case think is a semantic primitive as Wierzbicka (1972, 1980) demonstrates. The two arguments have semantic properties predetermined by this semantic structure of the predicate – x has the role of an experiencer and is a person or anthropomorphous being, whereas y (something.be.in.mind) may have the following semantic varieties:

1. perception(s) – something.x.perceived.be.in.mind

2. knowledge – something.x.knows.be.in.mind

3. belief(s) – something.x.believes.be.in.mind

4. intention(s) – something.x.intends.be.in.mind (Van Valin – Wilkins, 1993, p. 513).

It should be added that in the semantic structure of the three indicated verbs there is a union of three cognitive states:

[206]1. an initial state in which a real situation p is reflected in the mind, respectively a presumption of a possible situation p arises, or an intention of realization of a possible situation;

2. a state in which the situation p, which is a ‘product’ of the cognitive state (1), is stored in the mind;

3. a later state where the situation p is again activated in the mind. This situation p can be reduced to one object.

In the sentence, only the states (2) and (3) are expressed explicitly. The ‘hidden’ cognitive state (1), however, proves to be grammatically relevant since it is this state that defines the essential syntactic differences in the form of the argument y.


I. When y is an object of perception y may be a concrete or situation.

When y is a concrete, y may be viewed as an aggregate of the implicit features it possesses, otherwise one or more of its features are indicated in the sentence. In Bulgarian and Russian we have the following sentence models:

1.a) Bulgarian x pomni, spomnja si, spomni si z


Ti pomniš li moreto i mašinite … (Vapcarov).


Az ošte go pomnja.


Ne pomnja nikogo / nikoj ot tjach..

1.b) Russian x pomnit, vspominajet, vspomnil z


Ja pomnju more.


Starik vdrug vspomnil mat’, otca, jego krasivogo žerebca, bol’šije derev’ja vo dvore, druz’ja v škole.


Ja ne pomnju jego materi.

2.a) Bulgarian x pomni, spomnja si, spomni si y, Q(y)


Az go pomnja bodăr i energičen.


Pomnja staricata sednala na praga.

2.b) Russian x pomnit, vspominajet, vspomnil y, Q(y)


Ja pomnju jego bodrym i energičnym.


Ja pomnju staruchu, sedjaščuju na poroge.

3.a) Bulgarian x pomni, spomnja si, spomni si y kato Q


Studentite pomnecha Ivan kato muzikant.

3.b) Russian x pomnit, vspominajet, vspomnil z kak (v kačestve) Q


Studenty pomnili Ivana kak muzykanta.


Vse pomnili jego v kačestve predsedatelja soveta.

The formal differences between Bulgarian and Russian are linked to the following general principles: The Bulgarian analytic noun system due to which y in (1) is always in a general form. However, when y is a personal pronoun, it is in the accusative case both in positive and negative sentences; the existence of an opposition between the genitive-accusative and the nominative according to the feature human : non-human concerning the masculine interrogative pronoun koj and the forms deri[207]ved from it. Whereas in Russian, being a synthetic language, y is in the accusative or genitive case depending on the opposition animate : inanimate and negative sentence : positive sentence.

In (2.a) y has the same properties in Bulgarian as in (1.a). Q is an adjective or participle agreeing in gender and number with y. In Russian in (2.b) y has the same characteristics as in (1.b), however, here Q is an adjective or participle agreeing in gender and number but not in case – Q is in the instrumental case (to indicate a transient feature), which is a clear sign of movement of Q from an other sentence. Compare (9) and


Ja pomnju jego – on byl bodrym i energičnym.

In (3) y in Bulgarian has the same qualities as in (1) and (2), whereas Q is a prepositional nominal phrase with a noun in the common form. In Russian it is in the accusative case after the preposition kak, which means, as does the Bulgarian preposition kato, ‘incapacity of’ or ‘during the time when y was Q’. In Russian there is also a model in which Q is presented through the phrase v kačestve + the genitive case. The model in which Q is connected to a noun in the instrumental case without a preposition is marginal and subject to restrictions in Russian. Compare:


Ja pomnju Mariju ješče učenicej and


* Ja pomnju Mariju učenicej.

After the simple sentence models we now come to the complex sentence models in which y is a subordinate clause denoting the situation with a proposition p. But before we get that far we should mention a model that is closely linked to (2). In this model y is an argument in the main clause and an obligatory element in the semantic structure of the subordinate clause, although it might not appear in the syntactic structure.

4.a) Bulgarian x pomni, spomnja si, spomni si y, Conj/Inter/Exlam/Rel Q (y, …)


Pomnja go, če beše bodăr i energičen.


Pomnja go, kato (kogato) beše mlad i energičen.


Majkata si spomnjaše Mina kak beše zastanala na praga


Az ja pomnja koga dojde na rabota i kakvo svărši.


Az si go spomnjam kak chubavo peješe.


Toj ja pomneše vinagi takava, kakvato beše na dvadeset godini.

4.b) Russian Ja pomnju, vspominaju, vspomnil y, Conj/ Inter-Rel/Exlam Q (y, …)


Ja pomnju jego, čto on byl bodr i energičen.


Vsje pomnili Mariju, kogda jeje vybrali starostoj.


Mat’ často vspominala svoju doč’, kak ona stoit na poroge


Ja pomnju jego, kak bodr i energičen on byl.


On pomnil jeje, kakoj ona byla v svoji dvadcat’ let.


On pomnil jeje vsegda takoj, kakoj ona byla v svoji dvadcat’ let.

The main difference between Bulgarian and Russian in this respect, is that most of the aforementioned varieties are more marginal in Russian than in Bulgarian. There are also differences in the interrogatives and relatives, and in that Bulgarian has two sets of sentence connectors – interrogative and relative, whereas Russian has only [208]one set that serves the functions of the two Bulgarian ones (compare (27) and (28)). The sentences containing the words če, čto are factive representatives as are the sentences with interrogative words in which these words serve only to reduce the information content of the reply to the question the speaker has in mind (see Mel’čuk – Žolkovskij, 1984, p. 620). Sentences containing kak (on the different meanings of this word and details about the semantics and forms of the other sentences see Nicolova, 1998) have a prototypical character used to denotate a situation that has been the object of perception.

The interrogative pronouns kak, kolko, kakăv in Bulgarian and kak, kakoj, skol’ko in Russian may also be exclamative intensifiers denoting a high degree of feature (qualitative or quantitative) and due to the communicational dependence of the subordinate clause this exclamation is quite weak. In Russian there are two possible formal varieties of the predicate in a subordinate clause if the predicate is an adjective – with a short nominative form or with a long instrumental form if the predicate is a long adjective. In this case the feature’s connection to a definite moment in the past is emphasized.

Marginal in both languages are sentences containing a temporal conjunction indicating concurrence of the action or of the result of the action in which y has taken part, with the first cognitive state – usually the perception of the experiencer x. Coreference of the direct object in the main clause with an argument from the subordinate clause is obligatory in all models. If this argument is the subject, it is usually a null subject when the topic of the informational structure in Bulgarian, and in Russian is normally denoted by a personal pronoun. But if it is not the subject, in both languages this argument is designated by a personal pronoun with or without a preposition.

The most common type of complex sentences with the aforementioned predicates are accompanied by subordinate object clauses which denote the situation through the proposition p. In this case coreference of x with an argument from the subordinate clause is possible, see (33) and (36).

5.a) Bulgarian x pomni Conj/ Inter/ Exlam p/ Rel p


Pomnja, če Ivan živeeše do nas.


Pomnim kogato čovek stăpi na lunata.


Momičeto si spomni kak se igraeše tozi tanc.


Pomnja kolko bodăr i energičen beše toj.


Toj pomneše vsičko, koeto beše vidjal togava.

5.b) Russian x pomnit, vspominajet, vspomnit Conj/ Inter-Rel/Exlam p


Pomnju, čto Ivan žil do nas.


Devuška vspomnila kak tancujetsja etot tanec.


On pomnil vsje, čto togda uvidel.


Ja pomnju kak (naskol’ko, kakim) bodrym i energičnym on byl / kak, naskol’ko, kakoj bodr i energičen on byl.

The subordinate clauses in the model are the same as in (3.b).

The subordinate clauses denoting the situation which is the object of perception, may represent the attitude of different people towards the situation:

[209]a) of a person who is speaking about a memory of a situation he himself has experienced, viz. the speaker is also an epistemical subject (ES);

b) of a person who is speaking about an other person’s memory, i.e. the speaker is not ES.

Apart from this, the complex sentence can be direct speech when the speaker is the author of the words or renarrated speech when the speaker is conveying the assertions of others.

The simplest case is (a). Here, in Bulgarian, one uses vouched for tenses (the aorist, imperfect, pluperfect) and the praesens historicum, which requires the indication of at least two actions.


Pomnja kak profesorăt vliza veličestveno v auditorijata, pozdravjava ni i ni gleda strogo.

In Russian, the past tense and the preasens historicum are used.

If the speaker is not coreferential with ES, we have the two following possibilities:

a) the speaker has also witnessed a situation that an other person recalls and speaks about, or he has not witnessed it, but wants the witness’ attitude of ES towards p to remain unchanged, as in oratio obliqua in artistic speech. In this case, in Bulgarian one uses the aforementioned indicative tenses in subordinate clauses.


Ivan pomni kak vojnicite razrušicha (aorist) mosta.

b) the speaker indicates that he has not been a witness of p. In this case there are two possibilities in Bulgarian:

1. If the speaker is conveying information that has already entered his cognition, viz. he is rendering his own account of the veracity of the information, then the perfect tense of the indicative is used in the subordinate clause.


Ivan pomni kak vojnicite sa razrušili mosta.

Since the point of reference of the perfect is simultaneous with the point of speech, the perfect has null value in the ‘evidential’ feature of the action, because a person can be witness to an activity only in a defined moment of time in the past. This is why the perfect is particularly suitable for conveying information about a situation that has not been the speaker’s object of perception.

2. If the speaker is conveying information received from another person and wants to indicate this, in Bulgarian he would use the renarrative or the dubitative in the subordinate clause to express doubt about the truthfulness of p. It is interesting that the verb in the main clause is in the indicative if the speaker is communicating the speech of ES, and in the renarrative or dubitative if not transmitting the words of ES, but of a third person.


Ivan pomni kak vojnicite razrušili mosta.


Ivan pomnel (bil pomnel) kak vojnicite razrušili (bili razrušili) mosta.

In a broad sense, modal attitudes in Russian are not grammaticalized. Russian has neither modi loquendi nor tenses with a past point of reference. It can only be indicated lexically that the speaker is conveying someone else’s words. In Bulgarian, examples (39), (40) and (41) correspond to


Ivan pomnit, kak soldaty razrušili most, and to (42) –


So slov drugich, ja znaju, čto Ivan (mol, deskat’) pomnit, kak soldaty razrušili most.


II. When the second argument y of the predicates of recollection is the situation p which is stored in the mind as knowledge resulting from the mental activity of x, we have the following sentence models:

1.a) Bulgarian x pomni, spomnja si, spomni si za y


Momčeto si spomni za ključa.

1.b) In Russian x pomnit, vspominajet, vspomnil o/pro y


Mal’čik vspomnil pro ključ / o ključe.

This kind of simple sentences is summarized speech. On a semantic level, these sentences are incomplete since only one element of p is mentioned. For example


Momčeto si spomni, če ključăt e u majka mu.

In Bulgarian, this element of p is designated by a nominal phrase with the preposition za (usually with the definite article) and in Russian – by the synonymous nominal phrases with the preposition o + the locative or pro + the accusative. In order for the hearer to interpret this kind of sentences accurately, he must have preliminary information at his disposal.

2.a) Bulgarian x pomni, spomnja si, spomni si y


Učitelkata si spomni rešenieto na zadačata.

2.b) Russian x pomnit, vspominajet, vspomnil y


Učitel’nica vspomnila rešenije zadači.

In this model y is a noun or nominal phrase normally denoting the product of cognitive activity – a text, law, lesson, theorem etc. In Bulgarian, y is a noun in the common form, whereas in Russian it is in the accusative.

When y is a situation denoted by the proposition p, both languages have the following complex sentence models:

3.a) Bulgarian x pomni, spomnja si, spomni si Conj/ Inter/ Rel p


Ot urocite po chimija toj pomneše samo, če molekulata na vodata e săstavena ot vodorod i kislorod.


Ivan si spomni kăde se namira Angola.


Toj pomni samo (tova), koeto mu e neobchodimo za rabotata.

3.b) Russian x pomnit, vspominajet, vspomnil Conj/Inter- Rel p


Iz urokov po chimii on pomnil tol’ko, čto molekula vody sostojit iz vodoroda i kisloroda.


Ivan vspomnil, gde nachoditsja Angola.


On pomnit tol’ko to, čto jemu neobchodimo dlja raboty.

The complex sentences have the same qualities as the corresponding sentences in I (3.b).


III. The constructions in which y is an object of opinion have the following models:

1.a) Bulgarian x ne pomni, ne si spomnja, ne si je spomnil da p


Az ne si spomnjam da săm go sreštal.

 1.b) Russian x ne pomnit, čtoby p 


Ja ne pomnju, čtoby ja jego kogda-libo vstrečal.

These constructions are subject to two restrictions: the mental predicate can only be in the negative form (in the other semantic varieties of the constructions with mental predicates the mental predicate can be either in the positive or negative form) and its subordinate clauses with da in Bulgarian (with a verb in the perfect only) and the subordinate clauses with čtoby (with a verb in the past tense) in Russian, are connected to a non-factive presupposition. In Bulgarian, the complementizer da in subordinate clauses is a sign that these clauses imply a situation from some kind of possible world where the activity is possible, necessary or absolute (in other Slavic languages this is designated by the infinitive). We should note that subordinate da-clauses exist in Serbian, Croatian and Slovene, however, their meaning in these languages differ substantially from the meaning of the Bulgarian da-clauses.

As opposed to the Bulgarian da-clauses and the Russian čtoby-clauses, which are connected to the presupposition of non-factivity, the Bulgarian če-clauses, Russian čto-clauses and the sentences with interrogative pronouns are related to the presupposition of factivity regardless of whether the main clause is positive or negative. Only subordinate clauses containing the interrogative participles dali, li in Bulgarian and li in Russian have no relation to the presupposition of factivity or non-factivity, since their semantic structure is a disjunction: p or non-p.

In Bulgarian: 


Ivan ne pomni dali e izključil televizora ot kontakta.

In Russian: 


Ivan ne pomnit vyključil li televizor iz kontakta.

When being ES, the speaker may connect the proposition p at negation of the mental predicate to a factive or non-factive presupposition, but when the main clause is positive, the subordinate clause can only be linked to a non-factive presupposition.

In Bulgarian: 


Az pomnja, če profesorăt beše / e bil prijatel na Stojanov.

In Russian: 


Ja pomnju, čto professor byl drugom Stojanova.

Presupposition: I know that p.

Assertion: I claim that I remember that p, i.e. I find in my cognition that p.

When there is a negation of the mental predicate, the presupposition of factivity is also possible, although such sentences are rare since there is a certain inconsistency between the meaning of the main and subordinate clause. The use of such clauses is subject to defined pragmatic conditions: for example, the speaker has already received the information from the hearer that p, which he considers to be the truth and at [212]the same time claims that there is no information that p in his recollection. This means that there is a discrepancy between the external and internal information. 


Az ne pomnja, če profesorăt e bil prijatel na Stojanov.

Presupposition: I know that p.

Assertion: I claim that I do not remember that p, i.e. I cannot find the information that p is a fact in my cognition.

There is no inconsistency between the presupposition and assertion when in a negation of the mental predicate, the presupposition of the subordinate da- respectively čtoby-clause is non-factive. 


Az ne pomnja profesorăt da e bil njakoga prijatel na Stojanov.


Ja ne pomnju, čtoby professor kogda-nibud’ byl drugom Stojanova.

Presupposition: I think that M p (p is possible).

Assertion: I claim that I do not remember that p, i.e. I cannot find the information that the potential p is a fact in my cognition.

When the speaker and ES are not coreferential, possible sentences with the two types of presupposition at negation of the mental predicate and with the presupposition of factivity are in Bulgarian, if the mental predicate is in the positive form. Since the attitudes of two individuals toward p is involved, these attitudes may be identical or different. 


Ivan pomni, če profesorăt e bil prijatel na Stojanov.

Presupposition: I know that p.

Assertion: I claim, that Ivan knows that p – the speaker and Ivan have some information that p. 


Ivan ne pomni, če profesorăt e bil prijatel na Stojanov.

Presupposition: I know that p.

Assertion: I claim that Ivan does not remember that p – the speaker has the information that p, but Ivan does not have this information. 


Ivan ne pomni profesorăt da e bil prijatel na Stojanov.

Presupposition: I think that Mp (p is possible).

Assertion: I claim that Ivan does not remember that p, i.e. he does not find the information that the possible p is a fact in his cognition.

What has been said about the semantic constructions with the verb pomnja is perfectly valid also for the verbs spomnjam si, spomnja si as well as for several other mental predicates in Bulgarian. In Russian, however, the picture is different. When using the verb pomnit’ with a negation, the complement is usually a subordinate čtoby-clause with a presupposition of non-factivity (see (69)). It is possible to use a subordinate clause with čto or an interrogative pronoun only if the meaning of the subordinate clause doesn’t allow the non-factive presupposition. 


Ivan ne pomnit, čtoby professor byl drugom Stojanova.

Presupposition: I think that Mp (p is possible).

[213]Assertion: I claim that Ivan does not remember that p, i.e. he does not find the information that the possible p is a fact in his cognition. 


Ivan ne pomnit, čto ja jemu dal knigu.

Presupposition: I know that p.

Assertion: I claim that Ivan does not remember that p, i.e. he does not find the information that p in his cognition.

The verbs vspominat’, vspomnit’ with a negation are not combined with subordinate čtoby-clauses, but only with factive complements: clauses with čto and interrogative words. 


Ivan ne vspomnil / ne vspominal, čto professor byl drugom Stojanova.

Since these two verbs have different lexical meanings from the verb pomnit’, with the feature ‘resultivity’, which according to Padučeva (1996, p. 20–21) is inherent in the meaning of vspomnit’ and is implied in the meaning of vspominat’, one may put forward the hypothesis that in the presence of a negation, this feature correlates with the factivity presupposition of the subordinate clause and that the unmarkedness of this feature in pomnit’ correlates with the presupposition of the non-factive complement by also permitting a presupposition of the factive complement. It should be ascertained whether this correlation between a feature of the predicate’s lexical meaning and a modal feature of its complement’s presupposition is characteristic only of the words analyzed or if it is also found in constructions with other mental predicates.


IV. The cases where the complement of the analyzed predicate is a psych-action according to Van Valin – Wilkins’ terminology have been clearly distinguished formally. In these cases we have the following varieties:

1.a) Bulgarian pomni, spomni si, spomnjaj si Conj/Inter p 


Pomni (da pomniš) da ne izključvaš aparata ot kontakta.


Pomni (da pomniš) če ne trjabva da izključvaš aparata ot kontakta.


Spomni si kakvo e neobchodimo da napraviš pri požar.

1.b) Russian pomni, vspomni, vspominaj Conj/Inter p 


Pomni, čto ne nado vyključat’ apparat iz kontakta.


Vspomni, čto ty dolžen zavtra v pjat’ časov pozvonit’ po telefonu.


Vspominaj často, kodga nado pozvonit’ načal’niku.

In these examples the mental verb is in the imperative. In Bulgarian it may also be in the da form (on the difference between the imperative and the da-forms, see Nicolova, 1984). In this example the source of intention is the speaker, whereas ES must complete the action indicated in the complement. In Bulgarian, this complement is represented by a da-clause with a verb in the present tense or by a če-clause with a verb in various tenses and an obligatory modal operator of necessity (trjabva, ne biva, neobchodimo e, naložitelno e and others) which eliminates the modal factivity operator če, as well as by interrogative clauses with a modal operator of necessity (e.g. trjabva). In Russian, the complement is represented by a clause with čto or by [214]interrogative words in which there is a modal operator of necessity (dolžen, nado, nužno) that can be linked to a verb in the infinitive.

It should be noted that the verbs pomnja, pomnit’ on the one hand and spomnja si, vspomnit’, spomnjam si, vspominat’ on the other hand have different meanings: the former expresses: ‘store in your memory the information that you must do this and that’, whereas the latter means: ‘extract from your memory the information that you must do this and that’. These mental verbs in the imperative may have complements that do not denote an intention, but describe a particular situation which should become or is the object of knowledge. 


Pomni, če săbranieto e utre večer.


Vspomni, kogda u Maši den’ roždenija!

2.a) Bulgarian x pomni, spomnja si, spomni si Conj/ Inter p 


Az pomnja, če trjabva da ti vărna knigata v petăk.


Činovnikăt pomneše dobre koga (trjabva) da se obadi na načalnika si.

2.b) Russian x pomnit, vspominajet, vspomnil Conj/ Inter p 


Ja pomnju, čto ja dolžna vozvratit’ tebe knigu v pjatnicu.


Činovnik chorošo pomnil, kogda nužno pozvonit’ načal’niku.

The complements possess the same qualities as in the preceding models with the only difference that here the da-clauses are hardly used at all.

3.a) Bulgarian x pomni, spomnja si, spomni si za y

3.b) Russian x pomnit, vspominajet, vspomnil pro/o y

These simple sentence models represent summarized speech and are identical to II.1.a and II.1.b (see examples there).

In English, the complements of the verb remember have the following distribution: when y is the object of perception, the complement is in the gerund, when y is the object of knowledge or belief, the complement is a subordinate clause with that, and when y is a psych-action – the complement is an infinitive construction. In Bulgarian and Russian, the complements are more multiform formally than in English.

When y is the object of perception, we have simple sentences of the form x pomni y, x pomni y Q(y) and complex sentences with subordinate clauses with the conjunctions če, čto in Bulgarian and Russian, with interrogative and relative pronouns and adverbs in Bulgarian, whereas in Russian, with interrogative-relative pronouns and adverbs, with exclamative interrogative pronouns-intensifiers, with temporal conjunctions kato, kogato, kogda.

When y is an object of cognition, simple sentences are of the model x pomni za y, x pomnit pro/o y (in summarized speech) and x pomni y (with lexical constraints for y), whereas complex sentences have subordinate clauses with the conjunctions če, čto with interrogative and relative pronouns and adverbs in Bulgarian, and with interrogative-relative pronouns and adverbs in Russian.

When y is an object of belief, or more precisely of propositional attitude, there exist only complex sentence models in which the mental predicate in the main clause is ex[215]clusively in the negative form, whereas the subordinate clauses with da (and a verb in the perfect tense) in Bulgarian and with čtoby (and a verb in the past tense) only in combination with pomnit’ in Russian. These subordinate clauses are connected to a presupposition of non-factivity unlike the subordinate clauses in which p is the object of perception or cognition, which are connected to a presupposition of factivity.

When y is a psych-action, in Bulgarian the complements are da-clauses with a verb in the present tense, če-clauses or clauses with interrogative words with a modal operator of necessity, whereas in Russian – clauses with čto and interrogatives pronouns in which there is a modal operator of necessity in combination with an infinitive. There also exists a variety with a simple sentence of the type x pomni za y, x pomnit o/pro y which is summarized speech as in the case where y is the object of cognition.

The present analysis proves the fundamental concept of RRG that the lexical meaning of the predicates in combination with an independent multitude of morphosyntactic principles predict the syntactic structure of its arguments. In Bulgarian, this multitude of morphosyntactic principles include: the analytic noun system, the three article system, three-member case system of the personal pronouns, mere remnants of case-endings based on the feature human : non-human in the masculine singular of the interrogative pronouns and their derivatives, the presence of interrogative and relative pronouns, a nine-member temporal system in the indicative, the presence not only of modi actionis but also of modi loquendi (evidentials) etc. In Russian we have: a synthetic noun system, opposition between the features animate : inanimate in the nouns system, presence of interrogative-relative pronouns, long and short forms of the adjectives, three-member temporal system in the indicative, absence of modi loquendi etc.

With regard to decomposition of the lexical meaning of the predicate, we reach the interesting conclusion that the various semantic features or different combinations of semantic features are relevant not only to the syntactic structure of the arguments, but also to their semantics. For example, the choice of a noun phrase to represent the first argument x – the experiencer, which is the epistemical subject of the analyzed predicates, is influenced by the permanent element of the predicates’ meaning (think.again). The choice of the second argument – the theme is influenced by the variable element of the lexical predicates’ meaning.

For this reason, the following should be kept in mind with regards to semantic decomposition:

1. the relevance of a given semantic feature or combination of features to the semantic and syntactic structure of the sentence;

2. the semantic ‘agreement’ of this semantic feature or features with the lexical meaning and/or grammatical features of the argument and other elements of the sentence. In order to analyze the semantic agreement, one must make a semantic decomposition of both the arguments and the other elements of the sentence.

The limit of semantic decomposition of the lexical meaning is connected to the discovery of syntactically relevant semantic features (non-trivial semantic features in Apresjan’s terminology), which agree with the semantic features of the lexical and/or [216]grammatical meaning of other elements in the sentence, as well as with semantic features of the presupposition linked to the sentence. For example, the semantic feature ‘resultivity’ of the Russian predicates vspomnit’, vspominat’ agrees with the semantic feature ‘factivity’ of the presupposition of their complement even when there is a negation of the mental predicate. Thus, the postulation that there exists some kind of semantic atoms or primitives of decomposition or paraphrase seems unfounded for the given purposes.

The methodology proposed by RRG for examining the interrelation between the lexis and syntax is still inchoate and needs to be supplemented with the following components:

1. a component for the semantic and formal agreement and interdependence between the different arguments of one predicate;

2. a component for the semantic ‘agreement’ of the predicate with the non-arguments, the apposition words and other elements of the sentence (Apresjan and colleagues have already taken the first steps in this analysis);

3. a pragmatic component.

In order to demonstrate the vast significance of the pragmatic component, we may put forth the Bulgarian language, which in this field has the highest degree of grammaticalization of the Slavic languages. In Bulgarian, the cognitive states of the speaker, such as perception, cognition, belief, reception of information from others, getting hypothetical information through inference, transition accompanied by astonishment, going from ignorance to knowledge (by aid of the admirative), transition from knowledge to ignorance owing to forgetfulness (by aid of the transpositively utilized tenses of the indicative, namely the imperfect, pluperfect and futurum praeteriti) and others are grammatically designated. It is not possible to present the semantics of a number of complementary subordinate clauses without describing the semantics of the presuppositions related to them without describing their modality, temporal relations with the main clause and so forth.

The integral description of language has entered a new stage in the description of language on a syntagmatical level where, as on the level of paradigmatics, ‘tout se tient’. This places a myriad of entirely new problems in the foreground, the solving of which demands a new complex methodology. In this regard, Slavic studies, since the time of the Prague Linguistic Circle, has played a leading role which due to it being little known among Western linguists still has not received the international merit it deserves.




APRESJAN, Ju. D.: Integral’noje opisanije jazyka i sistemnaja leksikografija. Izbrannyje trudy, tom II. Moskva 1995.

BOGUSŁAWSKI, A.: You can never know that you know. Semantikos, vol. 10, 1988, p. 1–44.

BOGUSŁAWSKI, A.: Do świata przez język. Przegląd humanistyczny, 1997, p. 103–129.

DANEŠ, F.: Sémantická struktura větného vzorce. In: Otázky slovanské syntaxe, II. Brno 1968, p. 45–49.

[217]DANEŠ, F. – HLAVSA, Z. – KOŘENSKÝ, J. a kol.: Práce o sémantické struktuře věty. Přehled a kritický rozbor. Praha 1973.

DIXON, R. M. W.: A method of semantic description. In: D. Steinberg – L. Jakobovits (ed.), Semantics. Cambridge 1971, p. 436–474.

DOWTY, D.: Word Meaning and Montague Grammar. Dordrecht 1979.

JAKOBSON, R.: Beiträge zur allgemeinen Kasuslehre. TCLP, 6, 1936.

JAKOBSON, R.: Selected Writings. Mouton, The Hague – Paris – New York 1962–1988.

KAROLAK, S.: Składnia vyrażeń predykatywnych. In: Gramatyka wspólczesnego języka polskiego. Warszawa 1984, p. 11–121.

KORYTKOWSKA, M.: Typy pozicji predykatywno-argumentowych. In: Gramatyka konfrontatywna bułgarsko-polska, t. 5, cz. 1. Warszawa 1992.

MEĽČUK, I. A.: Opyt teorii lingvističeskoj modeli „smysl – tekst“. Moskva 1974.

MEĽČUK, I. A. – Žolkovskij, A. K.: Slovar’ russkogo jazyka. Opyty semantiko-sintaksičeskogo opisanija russkoj leksiki. Wiener Slavistischer Almanach. Sonderband, 14. Wien 1984.

NICOLOVA, R.: Pragmatičen aspekt na izrečenieto v bălgarskija knižovan ezik. Sofija 1984.

Nicolova, R.: Mjastoto na leksikata v gramatikata na roljata i referentnostta. In: Slavjanska filologija, t. 22. Dokladi za XII meždunaroden kongres na slavistite Kraków 1998. Sofija 1998.

PADUČEVA, J. V.: Semantičeskije issledovanija (Semantika vremeni i vida v russkom jazyke. Semantika narrativna). Moskva 1996.

STANIŠEVA, D.: Upotreblenije glagolov dviženija s učetom faktora adresata v bolgarskom jazyke. Zeitschrift für Slawistik, B. 30, 1985.

STANIŠEVA, D.: O lingvističeskom nasledstve Romana Jakobsona (K stoletiju so dnja roždenija). Contrastive Linguistics, 21, 1996.

VAN VALIN, R. D. – WILKINS, D. P.: Predicting syntactic structure from semantic representations: remember in English and its equivalents in Mparntwe Arrernte. In: R. D. Van Valin (ed.), Advances in Role and Reference Grammar. Amsterdam – Philadelphia 1993.

VENDLER, Z.: Linguistics in Philosophy. Ithaca 1967.

WIERZBICKA, A.: Semantic Primitives. Frankfurt am Main 1972.



Základní problémy integrálního popisu jazyka

Integrální popis jazyka typu RRG (Role and Reference Grammar) představuje nový přístup k deskripci přirozeného jazyka, předpokládající: 1. systematický zřetel k interdependenci predikátů a jejich argumentů, 2. zřetel ke vztahům mezi predikáty a neargumentovými prvky výpovědi, 3. zřetel k pragmatické dimenzi výpovědi. Tento přístup přináší velké množství nových problémů, jejichž řešení vyžaduje i novou metodologii. Právě v tomto ohledu hraje slavistické zkoumání již od časů Pražského lingvistického kroužku významnou roli, jíž zatím neodpovídá pozornost, kterou jí „západní“ jazykověda věnuje.

Sofijski universitet sv. Kliment Ochridski
bul. Car Osvoboditel 15, 1040 Sofija, Bălgarija

Slovo a slovesnost, ročník 61 (2000), číslo 3, s. 203-217

Předchozí Jitka Šonková: Mluvená čeština a korpusová lingvistika

Následující František Štícha: Nad velkou Gramatikou německého jazyka