In a between-subjects design, we study how chains of Iterated Reproduction of stories (e.g., Bartlett 1932, Allport & Postman 1947) may be influenced by the pragmatic purpose of reproduction. Iterated Reproduction chains start with one out of N seeds, i.e., stories composed of existing text material describing (anonymized) actual events, such as from newspapers or reports. Seeds cover potentially scandalous events, such as sexual harassment, fraud, blackmailing, manipulation etc. Participants in a single chain initially read a seed story and are then told to reproduce it for some pragmatic purpose. The reproduction they produce will be the story for the next generation of participants, and so on for M generations. Each seed will be fed into K parallel iterated reproduction chains.
One group of participants is told to reproduce the relevant facts as faithfully as possible using a cover story that the participant is an apprentice in a lawyer’s office who needs to retell the story to their boss. A second group of participants operates on a cover story according to which the participant hears the story at a social meeting (a party) and later retells the story to a different group of people at another informal gathering of friends. The second group is not told to be faithful but rather to try to be entertaining and to capture their audiences’ attention. A third group functions as a control group. They only receive the neutral instruction to retell the story without giving any specific context information.
We are interested in comparing several dependent measures in their temporal development and over the between-subject manipulation of pragmatic purpose. Every story of every chain is presented to an independent pool of L experimental participants who given subjective judgments along several dimensions. For example, we could ask:
- Based on the story you have read, how likely do you think MAIN CHARACTER would be convicted for NAME-OF-OFFENSE if the case was brought to trial?
- Based on the story you have read, how sympathetic do you find MAIN CHARACTER?
- Based on the story you have read, how strongly would you support that MAIN CHARACTER should be fired from her/his job?
- ...
Additionally, we will collect a number of objective variables for each story which are interesting from a linguistic point of view. For example, we could look at:
- The number of uncertainty markers, i.e., words or constructions that show that the story-teller is uncertain about some factual information, like probably, about five times or so, etc.
- The number of expressives, i.e., words or constructions that signal speaker-evaluation, like fucking, bloody, an insane amount etc.
- Use of intensifiers.
- The number of repetitions of factual information.
- Choice of lexical material (emotional content, strength (touching vs groping).
- …
- Construct seeds
- Write cover stories
- Implement Iterated Reproduction experiment (in Dallinger, using the Bartlett demo … ?)
- Run a pilot with one or two seeds, inspect responses
- Device dependent measures based on pilot data (subjective and objective)
- Set up subjective-judgements task and, if suitable, pilot on pilot data from iterated reproduction
- Fix statistical analyses, write analysis scripts based on pilot data and its derived measures
- Pilot other seeds with small
- Submit preregistration report
- Run full exp.
- Fix N - number of seeds (3 -- 5)
- Fix M - number of generations
- Fix K - number of parallel chains for a given seed
- Fix L - number of subjects for subjective judgements