Composition
- Verbal Reports as Data (Ericsson & Simon, 1980)
- A Model of the Writing Process (Hayes & Flower, 1980)
- Empirical Tests of the Model
- Knowledge Telling and Knowledge Transforming
Verbal Reports as Data (Ericsson & Simon, 1980)
- Introspection
- Collecting Reliable "Think-Aloud" Protocols
Introspection
- In the early days of psychology, introspection was used to study the mind.
- Psychologists were trained to examine their own thoughts and experiences.
- The results were taken as explanations of the mind.
- Introspection fell into ill repute because individual psychologists with different
theories had different introspective experiences.
Collecting Reliable "Think-Aloud" Protocols
- Verbal reports must be treated as data rather than explanations.
- Verbal reports must be concurrent rather than retrospective.
- The researcher must demonstrate that "thinking aloud" does not interfere with
the primary task.
- The researcher must demonstrate a connection between the contents of the verbal
protocols and some additional, more objective, performance measure.
- Don’t trust the results of a verbal protocol analysis until you have converging
evidence to support it!
- Example: Fletcher (1986)
A Model of the Writing Process (Hayes & Flower, 1980)
- Importance
- The Task Environment
- The Writer’s Long-Term Memory
- Elements of the Writing Process
- Relations Among the Elements
- Questions
Importance
- It is difficult to study language production because we can’t control the
materials.
- Hayes & Flower (1980) were among the first cognitive psychologists to even try to
study composition.
- This work has probably had more impact on educational practice than anything else in
cognitive psychology.
- It has shifted the focus of instruction away from the product and toward the writing
process itself.
- This model is more of a "task analysis" than a real model.
The Task Environment
- The writing process is constrained by the task environment.
- This task environment includes:
- The Topic
- The Audience
- The Writer’s Goal
- The Text Produced So Far
The Writer’s Long-Term Memory
- The writer’s long-term memory is another important source of constraints.
- The important information in long-term memory includes:
- Knowledge of the Topic
- Knowledge of the Audience
- Stored Writing Plans
- Story Grammars
- The Inverted Pyramid
Elements of the Writing Process
- Planning
- Generating Ideas
- Organizing Ideas
- Goal Setting
- Translating
- Reviewing
- Reading (Controlled)
- Editing (Automatic)
Relations Among the Elements
- Editing and generating can interrupt the other processes.
- Goals determine the occurrence of reading, translating, organizing, and (persistent)
generating.
- Individual differences in goal setting are responsible for individual differences in
writing style.
Empirical Tests of the Model
- The Protocol
- Hypothesis 1: The form of the material should vary from section to section corresponding
to changes in process from section to section.
- Hypothesis 2: The content statements in the protocol should reflect the distribution of
processes predicted by the model.
- Hypothesis 3: Generating will be more persistent during section 1, when the goal is to
generate, than during sections 2 or 3 when it is not.
The Protocol
- The Data
- Scoring
- Sections and Goals
The Data
- The model was tested using data from a CMU faculty member who thought aloud while
writing a 1 page essay on his worries.
- The data included:
- 14 single-spaced, typed pages of verbal protocol
- 5 pages of notes
- the one page essay
Scoring
- The verbal protocol was divided into 458 segments.
- Each segment was categorized.
- Metacomments
- Comments about the writing process itself.
- "I’ll make a list of ideas now."
- Content Statements
- Statements that reflect the ongoing process.
- "That’s not the right word."
- Interjections
- The garbage category.
- "OK." "All right." "Um."
Sections and Goals
- The protocol was divided into 3 sections, each with a different writing goal.
- Section 1: Generating
- "And what I’ll do now is simply write down some random thoughts."
- "Organizing nothing yet."
- Section 2: Organizing
- "Now I think its time to go back, read over the material and elaborate on its
organization."
- Section 3: Translating
- "Let’s try and write something."
Hypothesis 1
- The form of the material should vary from section to section corresponding to changes in
process from section to section.
- Two judges rated rated each written sentence or note (95% reliability) on the following
dimensions.
- Is it a complete grammatical sentence? (Translating)
- Is it indented, alphabetized or numbered? (Organizing)
- Is it associated in the verbal protocol with an interrogative suggesting a search for
sentence completion? (Translating)
Hypothesis 2
- The content statements in the protocol should reflect the distribution of processes
predicted by the model.
- All content statements in the protocol were classified as reflecting generating,
organizing, translating, or editing (95% agreement).
Hypothesis 3
- Generating will be more persistent during section 1, when the goal is to generate, than
during sections 2 or 3 when it is not.
- Two raters judged whether each content idea triggered the next (96% agreement).
- Dependent measure was the chain length.
Questions
- What have Hayes & Flower Proven?
- Have they followed Ericsson & Simon’s advice for collecting reliable verbal
protocols?
- How could the effects of changes in the task environment or the writer’s long-term
memory be captured by this model?
Knowledge Telling and Knowledge Transforming
- Scardimalia & Bereiter’s (1987) Knowledge Telling and Knowledge Transforming
models suggest that there are two types of writing.
- Knowledge transforming is an active, problem solving process.
- Knowledge telling is a simple memory dump.
- Kids start out as knowledge tellers and become knowledge transformers over time.
- Knowledge transforming is good for learning.
- Hayes & Flower’s model is about knowledge transforming.
- Is there anything "between" knowledge telling and knowledge transforming?
The End!
|