Meeting w/ Philip - 9/30/2009

Finished: Demonstrated Experiment 2 prototype on Mechanical Turk

To Do:
1. Convert Experiment 2 data (English) into .csv file
2. Format HTML to more closely match Stephan's version of Experiment 2
3. Repeat steps 1-2 for Arabic -- use whole sentences, not extracted agent/patient.
4. Try some experiments to see psychologically satisfying pay rate.

9/16/2009 -- Meeting w/ Philip

1. Experimental Design Questions
All of the blocking goes away
- EXP1 XOR EXP2
- No ordering
- Debating distractor roles right now (still do all questions)
- Check randomization. If none, hard-code randomized & reverse randomized set

2. People will need IRB form
- scan
- subjects will read, click "I agree"
- email for form when coded Exp 2

3. Materials emailed
- Exp 2 may need manual fixing

4. Tentative timetable
Thanksgiving -- analyzed results
Nov. 1 -- Data to Analyze
Early Oct. -- Arabic
2-3 wks -- English cleanup / gathering data

5. Reinstalling ubuntu today

6. Set up MT account

9/07/2009 -- 09/11/2009

Title: Multi-Language Lexical Structures

Objectives: TBA

Task completed:
1. Discussed starting directives with Professor Resnik.
2. Met with Olivia to discuss Mechanical Turk. She warned me to be careful in considering design aspects before beginning to transport the WebExp experiments to Mechanical Turk.

Skills developed:
1. Saw a demo of Mechanical Turk's interface and saw sample data output.

Anticipated goals:
1. Get reacquainted with Experiment 2's format and design requirements.
2. Read through MT's available APIs
3. Try coding a sample question on MT to make sure that MT will keep the original required formatting.