Children's Story Authoring with Propp's Morphology

Sean Hammond

The slides are here: slides.
The controls (in your browser) are:

Here are the notes that go along with the slides:


Slide: Title Slide
Introduce myself.
Slide: Story Maps
For my PhD I designed, implemented and evaluated a story authoring tool for children.
Slide: Propp's Morphology
The tool is based on Propp's classic model of fairy tale narrative structure, Propp's morphology.
Slide: Story Writing Session
I evaluated the tool in usability studies with children and teachers, and in empirical studies of children's story writing in schools.
I found that children were able to use the tool to plan and write their own original stories.
Slide: Results

Through using the Story Maps application children were able to grasp abstract narrative concepts and apply them to their own writing; using the application improved some aspects of children's stories; and children reported that they enjoyed using the application and found that it was helpful to their story writing.

Feedback from the children suggests that they enjoyed using the tool to write stories and found it helpful, that they felt a greater sense of creative freedom and ownership over their stories, and that the tool gave them confidence in the quality of their stories.

Slide: Outline



Slide: Outline
Slide: Riley & Reedy
Slide: A Woodlouse's Adventure
Slide: The Little Flower
These are typical classroom story-planning worksheets. TODO: Limitations of these.
Slide: Adventure Author
Slide: Script Cards
These are existing 'interactive narrative' computer applications. TODO: What's missing from these.

TODO: slide about beginning, middle & end; conflict & resolution; clear and consistent roles; motivations.

Slide: Perrault's The Fairies (1)
Slide: Perrault's The Fairies (2)
Slide: Perrault's The Fairies (3)
Slide: Perrault's The Fairies (4)

This story is given as an example of what I mean by fairy tale narrative structure.

These are medium-independent, they are features or narrative itself, rather than features of interactive narrative as a medium.

So what we want is a high-level model of narrative that embodies knowledge about the overall structure of a narrative and the purpose of components such as characters, actions and motivations within the whole.

Propp's Morphology

Slide: Outline
This thesis proposes that Propp's morphology is a suitable model.
Slide: Vladimir Propp
Propp was a Russian folklorist who in the early 20th century carried out a study of the narrative structures of traditional Russian fairy tales.
Slide: Afanasyev
Propp studied thousands of fairy tales from Afanasyev's collection of Russian folk tales.
Slide: Morphology of the Folktale

Propp presented the results of his study in his book Morphology of the Folktale.

In Morphology Propp aimed to decompose tales into their component parts and study the relationsips of these components to each other and to the whole. The result was an abstract structural composition or "morphology" that Propp claims underlies the type of tales that he studied.

Slide: Propp's Functions (1)

Propp breaks down tales into units that he calls "functions". This is an example of one of Propp's functions, Arrival. This one function appears in many different forms in many different tales. All of the concrete details: the characters, the objects, the locations, and the ways in which the function is realised, change from one tale to the next. But the function always remains the same.

Slide: Propp's Functions (2)

Propp defined a function as an act of a character, defined from the point of view of its significance for the course of action.

Slide: Propp's Functions (3)

Propp identified 31 functions that he claimed accounted for the plots of thousands of Russian fairy tales.

Not every story contains every function, the functions always occur in the same order, no function excludes another.

Slide: Propp's Functions (4)

While all of the concrete details of the fairy tales vary widely, the functions are remarkably consistent.

"This explains the two-fold quality of the tale: its amazing multiformity, picturesqueness, and color, and on the other hand, its no less striking uniformity, its repetition."

Slide: Propp's Functions (5)

What I've investigated in my thesis is the application of Propp's morphology as a story construction kit, to be used by children to plan and write their own original stories that display the true fairy tale form, that are structurally complete and well-formed.

You construct a story by choosing what functions you want and then filling in the details. - You can plan the whole thing first and then write it. - Or you can be more spontaneous, selecting some elements and writing that part of your story, then picking some more elements and filling in those. - Teachers can vary the components that are available to choose from, beginning with just a few of the basic components, and building up to a much richer model of narrative with a large number of components and with several different types of component representing differetn narrative concepts.

Slide: Thesis Questions
  1. How can Propp's morphology be represented in a story authoring tool for children?
  2. Can children apply Propp's morphology to story writing?
  3. Does using Propp's morphology affect the quality of the stories written by children? If so, in what ways?

Children = 7-11 year-olds.

Exploratory Study

Slide: Outline
Slide: Exploratory Study (1)

This was a two-week study in which children wrote stories using paper 'Propp cards' and 'story maps', an early prototype of the planned computer application.

Wanted to verify that children could grasp Propp's morphology and its abstract narrative concepts, such as the concept of function, and that they could use it to write stories with Proppian narrative structures.

Wanted to try out Propp cards and story maps as an interface.

Wanted to get experience working with children and story writing.

Slide: A Propp Card

This is one of the Propp cards used in the study. In the top-left is the name of the Propp function, an illustration, and some instructions. The rest of the card is left blank for the participant to write that part of their story.

Slide: A Story Map

This is one of the story maps used in the study. It is a sequence of Propp cards, laid out on a table or board.

Slide: Research Question

Can children use Propp cards and story maps to write stories that realise the Proppian narrative structures that the Propp cards and story maps aim to represent?

Slide: Week One: Recasting

In the first week children created new stories from existing sample stories. They were read a sample Proppian tale, shown the story map for the sample tale, and were then asked to write their own stories following the same story map.

Slide: Week Two: Constructing

In the second week the same children returned, and this time they wrote stories following story maps that they constructed themselves by selecting which Propp cards they wanted to use.

Slide: Interpretation of Propp's Morphology

We had to develop a means of assessing the stories written by the participants to decide, in a consistent and reliable way, what sequence of Propp functions was realised by each story.

These sequences could then be compared to the sequences of functions represented by the story maps used, to measure how successfully the participants had followed the story maps.

For this purpose we developed a clear, consistent and concise interpretation of Propp's morphology and a procedure for analysing a story using the interpretation, deriving a sequence of Propp functions from the story.

Slide: Quantitative Results

The quantitative results showed that the children were able to follow the story maps with a high degree of accuracy.

Slide: Limitations

These quantitative results have some important limitations:

They concern Propp's functions only, and do not take into account the many other components of Propp's morphology.

They do not account for the originality of the stories, do not distinguish between highly creative applications of Propp's morphology and simple rewrites of the sample stories. Qualitative analysis of the stories is needed.

They do not account for the quality of the stories, or compare stories written with and without using Propp's morphology.

They do not measure learning, no comparison of stories written before and after using Propp's morphology. Did the children appropriate Proppian narrative concepts from their experience with the tool?

Slide: Thesis Questions

So at this point in the research, where are we with respect to the thesis questions?

Q1. We have confirmed that Propp cards and story maps are a successful representation of Propp's morphology for a story authoring tool. These basic interface elements, along with observations about how the children used them and what worked and what didn't, can now feed into the design of a computer application that will further answer Q1.

Q2. We have shown that children can apply Propp's morphology to story writing. They were able to grasp Propp's abstract narrative concepts, such as the concept of function, and apply these concepts to their own writing.

Q3. We have not made much progress on Q3 yet, there has been no comparison of stories written with and without using Propp's morphology.

The Story Maps Application

Slide: Outline

Story Maps is a computer story authoring application that I developed and evaluated for this PhD.

Its design is based on the exploratory study:

Requirements were drawn up, a prototype implemented and tested with teachers and children, and then the final version implemented and tested.

Slide: Planning View (1)
Slide: Planning View (2)

The planning view is the opening screen of Story Maps. You are presented with a collection of story cards to choose from in the green area at the top. You can inspect each card more closely by moving the mouse cursor over a card to see it in more detail. You drag-and-drop the story cards that you want to use and arrange them into a story map in the grey area below. You can enter a title for your story in the toolbar at the bottom of the window.

Slide: Propp Cards

Propp cards were created for each of Propp's functions. Child-friendly texts for the cards were created based on our interpretation of Propp and on feedback from teachers. Illustrations were created by designer and illustrator Raymond Yuen, based on the texts.

Slide: Zooming

Each card has a title, short description, and longer description, these are progressively revealed as the user 'zooms in' on the card.

Slide: Story Editor

Clicking on the Write your Story! button brings up the story editor. With your story map in view above, you use the text editor below to enter the text for your story. You have to fill in a part of your story for each story card in your story map. If you change your mind about one of the cards in your story map, you can go back to the planning view at any time by clicking the Go back to planning button.

Slide: Saving & Opening

You can save your story map to file and open a saved story map to continue working on it later.

Slide: Export & Print

You can export your story to HTML or print it.

Slide: Logging

The application automatically saves your story and keeps a log.

Slide: Story Maps URL

Story Maps is cross-platform and free software, you can download it from the website: .

Experimental Study

Slide: Outline
Slide: Experiment
Slide: Research Question
Slide: Hypotheses

The expectation was that using the story maps application would improve some aspects of the stories written but not other aspects:

We expected the plot structure of the stories to improve. Story Maps should encourage stories to have a clear build up, precipitating event, protagonist's goal, action and resolution, because this plot structure is embedded in the Propp cards and their sequence.

We expected character roles to improve. Story Maps should encourage stories to contain characters that act in clear and consistent roles, such as protagonist and antagonist, because these roles are embodied by the spheres of action contained in Propp's functions.

We expected the coherence of the stories to improve, because the Propp cards embody a logically and artistically sound sequence and flow of events.

We did not expect the story maps application to affect the description of characters or settings and their attributes or of emotions because characters, settings and their attributes are not represented as first-class objects in the application.

Slide: Research Design

The study used a pre-test post-test with control group research design. Each participant wrote three stories during the study, at pre-test, control or treatment condition and post-test.

For the pre-test all participants wrote stories using "crib sheet" that was created with the help of a teacher.

Each participant was then randomly assigned to either the treatment group or the control group. Treatment group participants created stories using the story maps application, while the control group participants used a typical classroom story planner.

Finally, all participants wrote a third story for post-test, using the same crib sheet as in the pre-test.

Slide: Pre-test

For the pre-test participants read a "crib sheet" of questions to keep in mind before writing their stories. On the back of the crib sheet were some story titles to help with ideas.

Slide: Treatment

For the treatment participants were given a 15-minute introduction to the story maps application and the narrative concepts it uses. They then used the application to plan their stories in the computers. Participants printed out their completed story maps, and used these to write their stories on paper.

Slide: Control

For the control condition participants planned their stories using a typical story planning framework, and then wrote them.

Slide: Post-test

The post-test was the same as the pre-test, but the suggested story titles on the back of the crib sheet were different.

Slide: Quantitative Measure of Narrative Structure

To analyse the stories a quantitative measure of narrative structure was developed using an independent set of stories written by children of the same age group.

The construct of narrative structure that was to be measured was defined, the available procedures and tools were reviewed, and an existing analytic rating scale was selected. This scale was trialed on an independent set of real-world data and problems were found, so the scale was significantly improved based on the trial data. An inter-rater reliability test of the improved scale was carried out, with good results.

The measure contains 11 scales such as the one shown, each measuring a different criteria. It records whether or not the story has a protagonist and antagonist, major and minor settings, a precipitating event and some causal build-up to the precipitating event, whether the protagonist has a goal and initiates actions in order to achieve the goal, whether the story has a resolution, and a few further details. Each scale is simple and unidimensional, the levels are clearly defined, and empirical examples from the trial data exemplify each level.

This scale can tell us both whether and in what ways two sets of stories differ; it is reasonably quick to apply and can be used by non-experts; it is reliable; it is valid (it ticks all of the boxes on our defined construct of narrative structure); and it is more sensitive than other available quantitative measures.

Slide: Quantitative Results

We found that the story maps application did improve the narrative structure of the stories in some of the ways that we expected, but for some aspects of narrative structure there was no room for improvement. Specifically, Story Maps increased the incidence of stories that contained an antagonist and increased the causal complexity of story events leading up to the precipitating event.

But these conclusions must be qualified. First, due to an unexpected difference between the control and treatment groups at pre-test, the rival hypothesis of regression toward the mean cannot be strongly refuted on the basis of the quantitative data. Second, insufficient evidence was found to conclude that any effect was retained at post-test.

As expected, we found no evidence that the story maps application affected descriptions or emotions in the stories.

Slide: Thesis Questions

Where does this leave is with respect to the thesis questions?

  1. The Story Maps application has been developed and tested in formative usability studies and in empirical studies of children's story authoring in school settings. The application is a successful representation of Propp's morphology in a story authoring tool for children.

The Story Maps application has limitations, including a lack of support for descriptions of characters and settings, and that only the most fundamental concepts from Propp's morphology are represented.

  1. The exploratory and experimental studies show that children can apply Propp's morphology to story writing.

Again, only the most basic concepts from Propp have been studied.

  1. The experimental study has compared stories written using a story authoring tool based on Propp's morphology to stories written using a typical story planner, and shown (with qualifications) that certain aspects of the narrative structures of the stories were improved.

Further qualitative analysis could tell us more about the quality of the stories, and abut how children have used Propp's concepts in their stories.

Further Work

Slide: Outline
Slide: Further Work (1)

Several areas for further work have been identified, including:

Qualitative analysis of the stories, using Propp's morphology. Stories would first be broken down into their Proppian components, and then systematic description would proceed from this analysis. Children's stories could be compared to each other and to traditional tales. The aim is a rich description of how children have made use of Propp's morphology in their stories, what they can do with it.

Further formalisation of Propp's morphology.

Integrating more of Propp's concepts into the Story Maps application, enriching the model of narrative structure embodied by the application.

Longer-term study of children's acquisition of fairy-tale narrative structure through Propp's morphology. Both quantitative and qualitative analysis would be used. Pre-test post-test with control group, but with an intervention several sessions long. The aim is to show learning, to show a significant effect at post-test.

Slide: Further Work(2)

An agenda for further study could be as follows:

Take one narrative concept from Propp's morphology at a time.

Carry out formative studies to find out how to formalise the concept, and how to integrate it into a story authoring application.

The result of these formative studies would be a new version of the story maps application with an additional narrative concept represented.

This new version of the story authoring application could then be used in experimental studies of children's story authoring. Can children acquire the new narrative concept and apply it successfully? If so, how do they use the new concept in their stories?

You would then take the next narrative concept from Propp's morphology and begin the cycle again.


Slide: Questions?