Сборник 2002




Andrej A. Kibrik

Institute of Linguistics, Russian Academy of Sciences



Vera I. Podlesskaya

Russian State University for the Humanities



Tat’jana M. Kal’kova

Russian State University for the Humanities



Alla O. Litvinenko

Department of Theoretical and Applied linguistics, Moscow State Uniuversity




Keywords: discourse, narrative, cognitive structure, neurosis, rhetorical structure


We are reporting a study of Russian narrative discourse conducted by a group of linguists and neurologists. A corpus of narrative discourse was obtained from a group of children and adolescents with a diagnosed neurosis, and from a group of their matched controls. The Night Dream Stories were elicited through a special procedure and analyzed with the help of the Rhetorical Structure Theory by Mann and Thompson. Neuroses are based on an inner conflict, for instance between the person's desires and received social norms. We claim that it is possible to find linguistic signals of an inner conflict in the rhetorical structure of discourse. There are two main features of the rhetorical structure that signal a neurosis: (a) the higher complexity of the discourse, in terms of the overall story length and the depth of the rhetorical structure; (b) the higher frequency of certain rhetorical relations, such as Antithesis, Evaluation, Background, etc. We also found that the neurotic stories are peculiar in their global structure, often having a complication that is not resolved within the dream. We suggest that that is an iconic manifestation of fundamental conflicts always present in the cognitive structure of a neurotic child and not finding a resolution.



  1. Introduction


The purpose of this study is to investigate discourse structures in children’s Night Dream Stories and identify differences between normal and neurotic children. The label Night Dream Stories is given to establish continuity with the famous “Pear stories” chosen in Chafe’s (ed.) monograph [1] as a model for studying the cognitive structure of discourse.

The study is based on a corpus comprising 129 spoken narratives of which 69 are from subjects with diagnosed neuroses and 60 are from neurologically intact controls. The subjects were children and adolescents with a diagnosed neurosis, aged 8-17, patients of the Neurological clinic, Moscow Medical Academy and their age matched controls, i.e. normal children. All subjects were Russian native speakers. A subject could participate in more than one storytelling session. Before the storytelling experiment, somatic, psychological and neurological conditions of the children in the study were examined in order to define the gravity and type of a neurosis. Selecting patients for the experiment and developing methodology for their neurological, paraclinical (EEG, echo-EEG, X-ray) and psychological examination was done by V.Golubev and E.Korabelnikova [2]. Patients with mental diseases and patients with organic lesions were excluded from the study. Questionnaire-based and experimental psychological methods of examination included measuring depression, reaction to frustration, level of aggression, level of anxiety and distinguishing types of personality (labile, epileptoid, hysteroid etc.). Neurological examination aimed at distinguishing syndromes, like insomnia, anorexia, phobias, migraine etc., and at identifying stages, phases and clinical types of neuroses, like neurasthenia, obsessive/compulsive neurosis, anxiety neurosis etc. Clinical systematization of neuroses was undertaken according to the “International statistical classification of diseases and injuries” (ICA-10).

The selected subjects were instructed to tell in as much detail as possible about the night dream he/she had. The Night Dream Stories were elicited and audiotaped immediately after waking up in the morning. Eliciting Night Dream Stories immediately after waking up in the morning is an effective way to get the most adequate “direct” symbolic reflection of children’s dreams.

The assumptions underlying this study include:

  • Neuroses are always based on some kind of intra-psychic conflict, for instance between the person’s desires and received social norms , or between the person's self-image and the way other people view her/him. Neurotic children are concentrated on attempting to understand how the world is organized, what is their place in the world, and what should be considered correct and normal.
  • This inner conflict can be observed not directly but only through some overt behavior.
  • Recalled night dreams are among the central realizations of the underlying conflict.
  • Night dreams stories are one of the kinds of overt behavior that, first, is very informative regarding the neurosis, and, second, can be studied objectively by linguistic methods.



  1. Hypotheses


  • A Night Dream Story is a typical narrative, in that it is a portion of discourse in which a speaker describes a set of events in the real or imagined world. As is well known from prior research on human dreams, the dreamer is always present in one’s dream, either as a protagonist, or as an observer. Hence, Night Dream Stories belong to one of the most well represented types of narratives, namely, to personal experience narratives. The uniqueness of Night Dream Stories as narratives is that the reported personal experience is of a very special type, namely, the content of a dream is reported in the story as belonging to the inner – mental and emotional – world of the narrator “objectified” through self-awareness.
  • Normal and neurotic children may produce significantly different discourse structures and such differences can be captured and measured through methods of linguistic analysis.
  • Rhetorical Structure Theory by Mann and Thompson, see e.g. [3], can be useful in achieving this goal.
  • It is possible to find linguistic signals of an inner conflict in the structure of discourse produced by neurotic children. In particular, we can expect that narratives produced by neurotic children incorporate a lot of backgrounding material including comments, assessments, rich descriptions of events and related non-sequential situations, we can also expect extensive use of evaluatives and causals.
  • If so, then discourse structure can shed light on the cognitive structures of narrators’ minds.



  1. Methods


3.1. Transcription


The corpus of stories has been transcribed in accordance with a specially designed system of discourse transcription. Each story was broken into elementary discourse units (lines) that are further used as elementary nodes in a rhetorical net. This required a significant adjustment of the existing discourse transcription formats to Russian spoken discourse, especially in determining the necessary degree of phonetic/prosodic detail of the transcript, and in determining  boundaries of minimal discourse units. On some details of discourse transcription see works by I.V. Ganzin [4, 5].


3.2. Diagramming rhetorical structure


The discourse interpretation method we are using is based on the Rhetorical Structure Theory (RST) developed by William Mann and Sandra Thompson [3]. We have RST-diagrammed all stories in corpus, and the analyses presented in the rest of this paper are based on those diagrams. The RST diagrams were presented in the format based on the so-called “RST tool” (developed by Mick O’Donnell, http://cirrus.dai.ed.ac.uk:8000/tool) internationally accepted by discourse analysts who share the RST approach. G. Bronnikov (Russian State University for the Humanities) refined the tool so that it accounts for deviations from the tree structure and can process discontinuos nodes. The tool was also adapted to deal with the Cyrillic alphabet.

A text may have more than one RST analysis. This can result from analytical errors, but much more important are the cases of inherent text ambiguity. In order to accommodate alternative analyses, each story of the corpus was independently analyzed by three analysts. Discrepancies between analyses were then discussed in order to choose the most adequate interpretation which served then as a basis for our quantitative and qualitative results.

Some comments on RST are in order.

  • The main advantage of RST is that it provides a unified method for representing relations between discourse segments at all levels. RST postulates a set of rhetorical relations that are applicable to discourse units of any size – from minimal ones (normally clauses) to maximal ones (immediate constituents of discourse, such as episodes in stories). The more central discourse unit is called nucleus, and the supporting unit a satellite. Relations may be multinuclear or link a nucleus and a satellite. At RST diagrams nuclei are represented under vertical lines and satellites at the ends of arcs.
  • We have supplemented the canonical set of rhetorical relations postulated by the authors of RST (including somewhat over 20 relations) with an approximately equal number of additional relations, such as Begin, End, Setting, Discord, and some others. Currently we have around 50 rhetorical relations on our list (see the list of relations in Table 1; relations added to the canonical set are marked with an asterisk in Table 1).
  • The fundamental idea of RST is that discourse segments are added by the speaker not randomly but in order to contribute to realization of his/her communicative intention. Thus RST allows us to reconstruct the cognitive processes in the speaker as s/he unfolds the underlying communicative intention into a rhetorical net.






Background-out *

Begin *



Circumstance-out *


Comparison-out *



Content *

Degree *


Elaboration-out *

Emotional-reaction *


End *


Evaluation-epistemic *

Evaluation-out *


False-start *

Frustration *

Headline *

Interference *

Interpretation *


Limit *




Nonvolitional-cause-out *


Nonvolitional-result-out *

Obstacle *


Problem *


Repair *

Repeat *


Result-in-observation *

Setting *


Source-cognitionis *

Source-out *

Source-perceptionis *

Source-verbi *

Stimulus *


Topic *



Volitional-result-out *




Consequence *


Discord *





Table 1. The current working list of rhetorical relations



  1. Typical examples of Night Dream Stories


(Notes: “N” followed by a number indicates the ordinal number of the text in the corpus of stories produced by neurotic subjects; “Z” followed by a figure marks the ordinal number of the text in the corpus of stories produced by subjects from the control group, the mnemonic for Z is the Russian word zdorovyj ‘healthy’. Two dots with a number in parantheses indicate a pause and its length in seconds.)


Typical normal story:        Z11 J.C., 12, female


Level of anxiety: low. Level of depression: low. Level of aggression: low.

Reactions: extrapunitive

Tempo: 80 words/min

Measurements of discourse complexity (detailed quantitative results see in sections 8-13 of the present report)

Story length (number of discourse units): 16

Number of relations: 12

Degree of branching (a ratio of number of relations to story length): 0,75

Story depth, maximum: 4

Story depth, mode: 3

Story depth, mean: 2,45

Story depth, median: 3


  1. My s klassom ..(1.8) poshli ..(1.1) vot ..(0.5) kuda-to.

My classmates and I went somewhere.

  1. ..(0.3) Zashli v dom,

     Entered a house,

  1. ..(1.2) i tam ..(0.2) byli stupen'ki ..(1.8) i voda.

     and there were steps and water there.

  1. ..(1.0) My stali na bol'shoj plot,

     We went onto a big raft

  1. ..(0.6) i pereexali na druguju storonu.

     and crossed to the other side.

  1. ..(1.5) Potom ..(1.4) my vyshli iz dveri.

     Then we exited the door.

  1. ..(0.8) Tam byla dver' ..(1.0) takaja zheltaja.

     There was a door there, a yellow one.

  1. ..(0.5) My otkryli ee,

     We opened it,

  1. ..(1.2) i vyshli.

     and walked out.

  1. ..(2.7) I my ..(0.5) okazalis' na ..(0.3) jarmarke.

And we found ourselves in a fair.

  1. ..(3.3) I tam ..(2.0) byli vsjakie igrushki prodavalis'.

And there, there were various toys on sale.

  1. ..(2.0) I tam ..(0.4) prodavcy byli zveri.

And there the salespeople were animals.

  1. ..(2.5) Potom ..(1.5) my voshli eshche v odnu dver'.

Then we entered one more door.

  1. ..(2.1) Tam byla tropinka.

There was a trail there.

  1. ..(0.5) My proshli po ètoj tropinke,

We walked along that trail

  1. i vyshli ..(0.8) v shkolu.

     and came to the school.


Typical neurotic story:      N8, O.B., 14, female


Obsessive-compulsive neurosis

Syndromes: obsession, asthenia, insomnia. Phase: decompensation.

Level of anxiety: high. Level of depression: high. Level of aggression: low.

Reactions: intrapunitive

Tempo: 160 words/min

Measurements of discourse complexity (detailed quantitative results see in sections 8-13 of the present report)

Story length (number of discourse units): 23

Number of relations: 22

Degree of branching (a ratio of number of relations to story length): 0,96

Story depth, maximum: 9

Story depth, mode: 8

Story depth, mean: 6,5

Story depth, median: 8


  1. Ja byla doma ..(0.3) s mamoj, ..(1.1) s bratom,

     I was at home with my mom, with my brother,

  1. ..(0.4) nu tam ..(0.3) kot mne eshche snilsja moj.

     well I dreamt about my cat too.

  1. ..(2.8 m) Dolgoe tam vremja snilos',

     For a long time I dreamt

  1. kak my prosto doma tam,

     how we were just at home

  1. delami zanimaemsja.

     doing various chores.

  1. ..(1.) Potom ..(0.2) chego-to ..(0.2) trevogu ja pochuvstvovala,

     Then for some reason I felt anxiety,

  1. vygljanula v okno,

     looked out of the window,

  1. u nashego pod'ezda pozharnaja mashina stoit.

     next to our entrance there was a fire engine.

  1. ..(0.3) Ja smotrju,

     I am looking,

  1. ottuda plamja tak ..(0.5) polyxaet.

     flames are coming out from there.

  1. ..(0.7) Vot i ..(0.4) tam ..(0.3) ne znala,

     Well, I did not know

  1. chto mne delat'.

what to do.

  1. ..(0.4) Papy net,

Dad was not there,

14 ..(0.3) nu ..(0.4) mne pochemu-to kazalos',

     and I felt for some reason

  1. chto ja dolzhna vse reshit',

     that I had to decide everything,

  1. ne znala,

     I did not know

  1. kak nam spastis'.

     how we should save ourselves.

  1. ..(0.4) To est' mozhno bylo by po lestnice sbezhat',

That is, we could run down the stairs,

  1. lifty uzhe ne mogli rabotat'.

the elevators could not work anymore.

  1. ..(0.2) No s mamoj_

     But with mom…

  1. u menja zhe mama bol'naja_

     my mom is ill, you know.

  1. ..(0.9) I ja muchalas' dolgo.

     And I suffered for a long time.

  1. ..(0.3) A potom prosnulas'.

     And then I woke up.


  1. RST-based measurements of discourse structure


Two types of measurable features of the rhetorical structure that can be viewed as signals of a neurosis:

  • discourse complexity in terms of story length, story depth and the degree of branching (sections 6‑8)
  • frequency of individual rhetorical relations (section 9)

The calculations were made separately for three age groups of subjects: younger children from 7 to 11 years old (27 texts from children with a neurosis / 36 texts from neurologically intact children), older children from 12 to 14 years old (30 / 20) and adolescents (12 / 4). In the latter group the sample was too small to get statistically reliable results, these texts were used only for qualitative conclusions. In all tables below, N marks the neurotic sample and Z is an abbreviation for the control sample in each age group.

Quantitative differences between neurotic and normal texts are more evident in the group of older children. Although the study of developmental differences in narrative competence was outside of the scope of the present study, we may suggest that the observed differences in age groups can be attributed to narrative development in children. Most developmental linguists agree that by age of 5, normally developing children have access to most of the morphology and syntax of their language. Nonetheless, the ability to produce both a semantically coherent and a linguistically cohesive narrative is a complex developmental task, and its acquisition continues through the school years. We may hypothesize then that, in the younger group, differences in discourse complexity and in the distribution of individual rhetorical relations are mainly influenced by the developmental differences in narrative competence, rather than by the neurological status of a child. To prove this hypothesis, however, we need further investigations, possibly including additional data, especially narratives elicited from older adolescents and from adults.



  1. Discourse complexity: Length


Story length was calculated as a number of elementary discourse units in the story (corresponds to the number of lines in transcripts). (The methodology for this and other measurements was originally developed by A.O. Litvinenko in [6].)



Younger (7-11)

Older (12-14)

Adolescents (15-17)








Story length, median







Story length, mode







Story length, mean








Table 2. Story length measurements


The distribution of this variable appeared to be close to normal, hence the t-test was applicable. It showed that neurotics tend to produce longer dream stories than normal children: the differences in length between the neurotic and the control group are very close to what can be considered significant (t=1,89, p<0,05 in the younger group, t=1,95, p<0,05 in the older group with t=2 being the critical value). Since the variance is rather high in all samples, it is very probable that the significance will increase if we increase the number of texts in samples. The same tendency is observed in the group of adolescents, but the sample is too small to be statistically reliable. A notable fact is that, in all age groups, dream stories longer than 40 units are produced almost exclusively by neurotic subjects.



  1. Discourse complexity: Depth


To measure the story depth, the depth of each node in the rhetorical structure was measured, calculated as the number of nodes between the given node and the zero (“root”) node, and then, basing on the individual nodes’, the maximum, the mean, the mode and the median of the overall story depth was calculated. Table 3 below demonstrates one of those measurements, namely maximum story depth. The comparison within age groups showed that neurotic children in the older (12-14) group produce significantly deeper stories than their age matched controls (t-test, p<0,01). In the group of adolescents, the sample is too small to be statistically reliable.



Younger (7-11)

Older (12-14)

Adolescents (15-17)

Maximum depth































Table 3. Maximum story depth measurements


  1. Discourse complexity: Branching


The degree of branching was calculated as a ratio of number of relations in a story to the story length (the number of discourse units in the story). We have introduced this measurement to reflect the “flatness” of the rhetorical structure: it shows the rate of multinuclear relations in a story capturing the proportion of asymmetric rhetorical relations that complicate the flat basic structure of a narrative. In our data, this measurement doesn’t follow the normal distribution and, therefore, the nonparametric Van der Varden test was applied. It showed significant (p<0,01) differences in branching for the group of older (12-14) children: the mean branching differs as 79,60 for normal children vs. 88,27 for neurotics, medians and modes differ as 80 (Z) vs. 89,5 (N) and 83,0 (Z) vs. 91,67 (N), respectively. As we expected, in this group, neurotic narrators produce more vertical, more branching stories than normal children, while younger children do not show significant differences. The group of adolescents is, again, too small for the quantitative comparison.










Adolescents (15-17)





Adolescents (15-17)

Total number of texts





























Table 4. Branching measurements


  1. Frequency of individual relations, frequency of discourse markers


We have compared frequencies of individual rhetorical relations calculated as a ratio of the number of tokens of a given relation in a sample to the total number of discourse units in the sample. In Table 5 ratios are given multiplied by 100 representing the frequency of a given relation per 100 discourse units in each group. The following relations showed significant differences in the older group of subjects:



Per 100 units
Per 100 units



















Emotional Reaction
















Table 5. Frequencies of individual relations in the older group


Table 5 shows that the two basic multinuclear relations responsible for the flatness of narrative structure, namely, Sequence and Consequence have lower frequencies in neurotic texts. This is in full accordance with the higher branching and depth in this group, since it correlates with higher frequencies of asymmetric nucleus-satellite relations.

The End relation is almost twice as less frequent in neurotic texts. The reason for that is that typical End satellites (like “That’s it”, “That’s what my dream was like: very interesting” etc.) clearly show the border between the world of the dream and the actual situation of narration which is more typical for normals. A similar explanation can be given for lower frequencies of the Source-out relation in neurotic texts: Source-out satellites are portions of texts like “I dreamt that…”, “I saw in my dream that…” etc. that usually appear close to the very beginning of the story or at the beginning of a new episode (to remind that we are still within the scope of a dream). The difference in frequencies here is, however, less significant than in case of the End relation. We suggest that, given that the storytelling task was to tell about one’s dream, the transition to the “inner” world of the dream (as well as reminding that we are still inside that “inner” world) is, in both groups, often inferred from the context and less necessary to mark than the opposite transition, i.e. returning to the real actual situation of narration.

An opposite tendency is shown by the Background-out and Evaluation-epistemic relations. Background-out satellites, like line 21 in N8 above, bring in portions of information about the real world incorporated into the dream story for better understanding of what is going “inside” the dream. Evaluation-epistemic satellites are portions of discourse that show the speaker’s doubts whether what is said in the nucleus correctly conceptualizes what she/he saw in the dream. A typical example is shown below:


N4, R.T., 10


  1. ..u menja byla noga bol’naja_

I had a painful leg

  1. nu…xromala_

[I] limped.

  1. ..nu .. perelom chto li?

Well, did I break it or something like that?

  1. Da. Perelom.

Yes. [I think] I broke it.


In the quoted fragment, line 23 is an Evaluation-epistemic satellite to line 24: it presents the narrator’s doubts on how to assess what happened to her leg – whether it was broken or perhaps injured in some other way. Her final conclusion – that the leg was indeed broken is given in the nucleus, that is line 24.

Neurotics tend to intensively connect and compare what they saw in the dream to what they really experienced or observed in real life. We conclude that this is the reason why the frequencies of the Background-out and Evaluation-epistemic relations is several times higher in neurotic texts.

The significant difference in representing the Source-cognitionis relation proves that neurotic children show greater concern for exposing their mental states and mental actions. A typical example are lines 14-15 and 16-18 (“and I felt for some reason that I had to decide everything” and “I didn’t know how we should save ourselves”) from text N8 cited above, where the nuclei present the content of the narrator’s doubts and beliefs. Similarly, neurotic children show greater concern in giving accounts for their decisions and actions, that is why the Justify relation is more frequently represented in neurotics’ stories, cf. lines 18-21and 13-15 in N8.

Table 5 also shows higher frequencies in presenting Emotional Reaction in neurotic texts, which is in full accordance with our expectations. A typical example is from N4: “there was, like, a bog [nucleus] and I got so frightened [Emotional Reaction satellite]!”

Finally, neurotics show higher frequencies in presenting adversative relations. Radical difference is observed in marking a nucleus-nucleus Discord relation and a nucleus-satellite Antithesis relation. The Discord relation links two situations that are equally important to discourse development, but are considered incompatible by the narrator. That can be illustrated by lines 19-21, text N10: “She probably thought that I have hidden myself behind the bush [nucleus] while I laid down on the ground a bit further, behind a tree [nucleus]”. The Antithesis relation links a satellite, which expresses a proposition which the narrator refuses to identify with, and a nucleus, which  presents a contrasting proposition which the narrator does identify with. That can be illustrated by lines 1-2, text N42: “I was not that game, you know, Play Station [satellite], but rather … well …I was a central figure of that game [nucleus]”. Both the Discord and the Antithesis relations are based on the speaker’s evaluation of norms and expectations and, hence, can be treated as signals of a higher concern of these norms and expectations.

Adversative relations in Russian belong to rhetorical relations that get regularly signaled by discourse markers. For example, out of 31 case of the Antithesis relation in our corpus, 20 are marked with the conjunction A, which, in its turn, is distributed as 3 to 17 tokens in normal vs. neurotic texts. Similarly, out of 34 cases of the Discord relation 16 are marked with the conjunction NO, which, in its turn, is distributed as 2 to 14 tokens in normal vs. neurotic texts. This opens the possibility to use the distribution of particular discourse markers as a supporting criteria in linguistic diagnostic tests of neuroses. See [7] for details.



  1. An additional comparison:

Global discourse structure, or narrative schemata


We complemented our RST analyses of Night Dream Stories by comparing their macrostructure attempting to identify genre-specific schemata of narratives. As is well known, the classical narrative schema comprises the following elements:

  • Schema A: beginning – setting – complication – climax – denouement – coda.

However, dream stories differ significantly from regular narratives in not having the triad “complication – climax – denouement”. In dream stories, the narrative nucleus typically consists of a sequence of narrative clauses without a clear internal structure:

  • Schema B: beginning – setting – narrative sequence – coda.

The majority of narratives by normal children follow schema B. A smaller fraction is produced in accordance with the classical schema A. However, in neurotic children’s stories the following schema is very typical:

  • Schema C: beginning – setting – complication – narrative sequence – coda.

In other words, neurotic stories typically contain a complication that is not resolved within the dream. This may be an iconic manifestation of fundamental conflicts always present in the cognitive structure of a neurotic child and not finding a resolution.

As shown in Table 6 below, normals’ stories demonstrate strong preference for schema B (typical nonsensical dream). In neurotic dream stories, all three schemata are met with roughly equal frequencies. Especially striking is that Schema C (complication without denouement) is more than three times more frequent in neurotic stories compared to normals. See [8] for further details.



Schema A

Schema B

Schema C



15 (25%)

39 (65%)

6 (10%)

60 (100%)


22 (32%)

24 (35%)

23 (33%)

69 (100%)


Table 6: Distribution of stories across schema types


Below are Night Dream Stories illustrating each of the schemata (only English translations)


Typical A Schema (the classical narrative schema): Z28 M.M., 8


  1. My dad and I, we drove around the town.
  2. Then we went out to have a rest.

3-4.      And suddenly I came to the car,

  1. and the car was not there.
  2. I went ahead to look for the car.
  3. Suddenly I saw a policeman
  4. and asked him
  5. how I can get home.

10-11.   He called home

12-13.   I got picked up

  1. and was taken home.
  2. That’s all.


Typical B Schema (typical nonsensical dream): Z41 A.M., 9


  1. I had a dream,
  2. that when I grew up
  3. I bought an apartment.
  4. Bought a car,
  5. passed the driving test
  6. and got a car,
  7. and then I got used to that,

8-9.      and I gave a lift to mom,

  1. we swam in a river with mom,

11-12. and then I went to some town for vacation,

  1. and then I came back here only in three years.
  2. And mom was very happy.
  3. That’s all.
  4. The end.


Typical C Schema (complication without denouement): N29 M.I., 8 Anxiety neurosis

  1. There are bars,
  2. and I am falling downwards,
  3. and there is a sharp peak at the bottom,
  4. and I am landing on it,
  5. and sit.
  6. There is forest in the front,
  7. it is somehow dark around,
  8. as if it were evening.
  9. Suddenly a huge spider crawls out of the forest.
  10. So huge
  11. that it seems it could trample the whole forest as a leaf.
  12. And it seems to approach me
  13. and wants to eat me up
  14. Then I woke up.



  1. Conclusions


  • Neurotics produce stories that are longer than the normals’ stories.
  • Neurotics produce stories that are deeper than those of normals.
  • Neurotic narrators produce more vertical, more branching stories while normal children produce more flat stories.
  • Neurotics demonstrate greater frequency of adversative situations in their stories, greater concern for indicating complex structure of events, causality, and correctness of verbalization. Normals demonstrate clearer distinction between the dream situation and reality.
  • Neurotic stories typically contain a complication that is not resolved within the dream. This may be an iconic manifestation of fundamental conflicts always present in the cognitive structure of a neurotic child and not finding a resolution.
  • GENERAL: hierarchical discourse structure is significantly different in normal and neurotic stories; discourse structure sheds light on the cognitive structure of narrators’ minds.





  1. Chafe, Wallace (ed.) The pear stories. Cognitive, cultural, and linguistic aspects of narrative production. Norwood: Ablex, 1980.
  2. Golubev V.L., E.A.Korabelnikova. “Osobennosti snovidenij pri nevroticheskix rasstrojstvax u detej i podrostkov” [Features of night dreams of children and adolescents with neurotic disorders]. Nevrologicheskij vestnik, 1-2, 2001
  3. Mann W.C., Matthiessen C., Thompson S.A.. “Rhetorical structure theory and text analysis”. In: Mann W.C., Thompson S.A. (eds.) Discourse description: Diverse linguistic analyses of a fund-raising text. Amsterdam: Benjamins, 1992, pp.39-78.
  4. Ganzin, Igor’ V. “Izuchenie snovidenij u detej s nevroticheskimi rasstrojstvami” [A study of night dreams in neurotic children]. Tavricheskij zhurnal psixiatrii, 4-2, 2000.
  5. Ganzin, Igor’ V. “Paralingvisticheskie xarakteristiki presentatsii snovidenij u detej s nevroticheskimi rasstrojstvami” [Paralinguistic features of presentation of dreams in children with neurotic disorders]. Tavricheskij zhurnal psixiatrii, 4-3, 2000
  6. Litvinenko A.O. 2000. Teorija ritoricheskoj struktury kak universal’nyj metod opisanija diskursa. [Rhetorical structure theory as a universal method of discourse description] M.A. thesis. Moscow State Univ.
  7. Podlesskaya Vera I, and Tat’jana M.Kal’kova. “The order of syntactic constituents vs. the order of discourse units: the case of Russian adversative constructions”. In: Bohumil Palek and Osamu Tsujimura (eds.) Item order: its variety and linguistic and phonetic consequences. Prague: Charles University Press, 2001.
  8. Kibrik, Andrej A. “Cognitive discourse analysis: Some results”. In: Cognition in Language Use. Selected papers from the 7th International Pragmatics Conference, Vol. 1. Edited by Enikö Németh. Antwerp, International Pragmatics Association, 2001, 164-180.





Когнитивная структура нарративного дискурса: анализ

детских рассказов о сновидениях

А. А. Кибрик, В. И. Подлесская, Т. М. Калькова, А. О. Литвиненко



Ключевые слова: дискурс, нарратив, когнитивная структура, невроз, риторическая структура


В докладе сообщается об исследовании русского нарративного дискурса, проводимом группой лингвистов и неврологов. Изучаемые нарративы – детские рассказы о сновидениях, собранные по специальной методике. Корпус рассказов состоит из двух частей: рассказов детей, больных неврозами, и рассказов здоровых детей из контрольной группы. Главным инструментом анализа является Теория риторической структуры, созданной У.Манном и С. Томпсон. С точки зрения риторической структуры, два основных признака рассказов невротиков – большая сложность дискурса, измеряемая определенными количественными параметрами, и большая частота некоторых риторических отношений, в особенности адверсивных. В исследовании используется также сопоставление двух частей корпуса с точки зрения глобальной структуры рассказов. Невротические рассказы характеризуются ярко выраженной кульминацией (отражение конфликта в сознании рассказчика) и отсутствием развязки.


[1] This study has been supported by grant 459/1999 of the Research Support Scheme (Open Society Foundation) and a research grant of the Russian Foundation for Basic Research ("Дискурсивный анализ "Рассказов о сновидениях": использование лингвистических методов в дигностике детских неврозов"). We express our gratitude to both foundations for their assistance.