VionColas Corpus

Monique Vion
Laboratoire Langage et Parole
Université de Provence

Annie Colas
Université de Provence

Participants: 191 children ages 7, 9, and 11
Type of Study: picture descriptions
Location: France
Media type: no longer available
DOI: doi:10.21415/T5N300

Browsable transcripts

Download transcripts

Citation information

In accordance with TalkBank rules, any use of data from this corpus must be accompanied by at least one of the above references.

Project Description

This corpus was designed to study the cognitive constraints (memory based and/or inferential) that affect the establishment and management of links between events. The general hypothesis was that the linguistic expressions that structure discourse are the manifestation of conceptual constraints imposed by the information management process. By varying the conditions of information availability, inference making, and thematic continuity in pictorial narratives (silent comic strips), we provided verbalizing conditions that were more or less favorable to establishing conceptual relationship.

Each comic strip contained eight frames (8 x 8 cm). The first frame showed two characters. All subsequent frames showed only one of the two characters carrying out various activities. A minimal link between the frames was achieved by the continuous presence of one of the characters from the first frame.

Four different comic strip versions were constructed using a factorial combination of two variables: thematic continuity and layout. The first variable concerned “thematic continuity”. In the maintained topic condition, the materials were designed in such a way that a topic would be induced after the first frame by the repeated presence of the same character in every frame, up to and including the last one. In the changed topic condition, the materials were designed in such a way that a thematic break was generated by the reintroduction in the last picture of the other character from the first frame (in other words, frame 1 had both characters, frames 2 through 7 showed only one of the two characters, and frame 8 showed only the other). The second variable was a secondary one used to control the layout of the characters in the frames. To avoid any bias in referent marking brought about by the greater salience of one of the two characters due to its location in the picture, the layout (left, right) of the characters in the first frame was counterbalanced.

The comic strips differed as to whether event sequence was arbitrary or ordered. In the arbitrary sequences, the events although presented as a sequence, could have occurred in any order (e.g., in A1, the daily activities depicted are relatively independent of each other, and thus required inference making: the woman getting dressed -or undressed- could have been placed after the women putting on -or taking off- her makeup, or anywhere else in the sequence, for that matter). In this case, the speaker's had to infer the links between the pictures from the proposed sequence in order to build an overall representation of one story. In the ordered sequences, the order of the events could not be changed (e.g., in O12, before potentially catching a fish, the man had to put on his fishing gear, go to the water's edge, and cast the line). The ordered sequences still did not have a script structure because the normal sequence of events was modified by the sudden appearance of an obstacle. The obstacle was always an event over which the main character had little or no control. In some of the comic strips, the obstacle interrupted the causal chain of events (e.g., in O15, the car hit a hedgehog crossing the road). In others, the obstacle did not interrupt the causal chain but created a surprise effect that sometimes substantially changed the expected course of events (e.g., in O16, the air bubble the fish entered so it could fly burst) and sometimes did not (e.g., in O13 the cereal bowl fell and made a hat for the cat hanging on the tablecloth). For each type of sequence, the materials consisted of 32 test comic strips (8 pairs of characters x 4 versions).

The last variable manipulated was the frame display mode. In the simultaneous display mode, all pictures were on one page. The speaker was asked to look at the comic strip and to prepare to tell the story immediately afterwards. In the consecutive display mode, the comic strip was presented in booklet format, with one picture per page. Subjects were instructed to turn the pages one by one and to say what was happening on each page. As such, the events had to be verbalized on-line, as they were discovered.

One hundred and ninety-one native French-speaking subjects (98 males and 93 females) participated in the study. There were 63 seven-year-old children (attending first grade), 64 nine-year-old children (attending third grade), and 64 eleven-year-old children (attending fifth grade).

Data collection design Each speaker was tested in only one frame display mode and on one type of sequence. During testing, a given participant saw eight test comic strips (each presented in one of the four versions).

Testing was individual and lasted approximately 20 minutes. In the room where the experiment took place, there were three persons, the speaker, the experimenter, and the addressee of the narration. The addressee was a same-age peer from the speaker's grade in school. He/she only acted as listener once during the experiment.

1528 narratives were audiotaped in several public elementary schools in Aix-en-Provence (Château-Double, Henri Wallon, Les Granettes) and Luynes (public elementary school and St François d'Assise) France. We would like to thank Delphine Baigue and Aïcha Idriss-Abdalla (graduate students at the time) for their help in preparing the materials and collecting the data.

The description of the picture stimuli is as follows. First, the arbitrary sequences:
A1: Un homme et une femme(A man and a woman)
A2: Un adolescent et un garçonnet(An adolescent and a little boy)
A3: Un homme et un adolescent(A man and an adolescent)
A4: Une femme et une fillette(A woman and a little girl)
A5: Une tortue et un crocodile(A tortoise and a crocodile)
A6: Un singe et un lion(A monkey and a lion)
A7: Une poule et des poussins(A hen and chicks)
A8: Un chat et un âne(A cat and a donkey)
And then the ordered sequences:
O9: Garçon et grand-père(Boy and grand-father)
O10: Homme et femme(Man and woman)
O11: Garçon et fille à la plage(Boy and girl at the beach
O12: Fils et père à la pêche(Son and father fishing)
O13: Chien et chat(Dog and cat)
O14: Ver et escargot(Worm and snail)
O15: Hérisson et lapin(Hedgedog and rabbit)
O16: Poisson et grenouille(Fish and frog)

File names are constructed using the first three number for the participant ID and the fourth and fifth for the age (07, 09 and 11). Then come three letters. The first two are either im (arbitrary) or ex (ordered) and the last is either g (simultaneous) or s (consecutive).

The gem codes in the files begin with V for vignette. Then there is a number for a picture number of the letter “u” for the picture series 2 through 7 and then the letter “m” for maintaining topic or “c” for changing topic. If the child went straight from the “u” sequence on to sequence 8 without a break, then the CHAT will include +… followed by +^ as in this example:

(end of VU) *CHI: après il joue du piano +...
(V8m) *CHI: +^ et il va se coucher à l'ombre.