Programming assignment 4
Overview
In this programming assignment you will work with raw production data. The goals are for you to have a better understanding of how to manage large projects and to gain experience analyzing stop consonants. You will generate an a priori hypothesis, segment audio files, label them with textgrids, use them to create a data set, and analyze said data set. To this end you will need to (partially) write/troubleshoot praat scripts. Once you have a data set, you will use your R skills to tidy the data and provide basic descriptive statistics and visualizations in a reproducible report. All steps/answers/etc. must be documented and reproducible.
Assigned: Week 11 (11/13)
Due: Monday, 11/25 before 10pm
About the experiment
You will fork a repo that includes a production experiment. Do not worry too much about the procedures involved in creating the recordings. Your focus is on learning to organize a project, extract data, etc. That said, you should know a little bit about the recordings. You will work with audio files from 6 participants: 3 bilinguals, and 3 L2 learners. All the participants were recorded saying nonsense words in Spanish (or what they believed to be Spanish). The target words are paroxytones that contain voiceless stops in disyllabic sequences. All 5 oral vowels of Spanish are included and each target was repeated 3 times. Think about this information and generate a hypothesis regarding what you think you will find (see below). The praat scripts you work with will extract voice onset time and the first two formants of the stressed vowel.
Procedure
Fork the repo here: https://github.com/jvcasillas/pa_4
See below if you don’t know what this means. Take time to carefully examine the structure of the repo. Many of the folders contain README text documents. You should pay special attention to the information contained in these files, as it is unlikely that you will be able to complete this assignment if you do not. Inside the report folder there is a README.Rmd file. This file is where you will conduct all of your work in R and respond to all questions. Read it carefully.
Submit
Your project will be a fork of my pa_4 repo. After you fork the repo, you should clone it to your computer using the github desktop app and work with it from there, i.e., locally on your computer at home. Commit and push all changes in your copy of the project to github using either the github app or RStudio. When you finish the assignment create a new issue and tag me (@jvcasillas). No need to submit a pull request. Make sure your fork of the repo is public! Note: I am interested in the entire project. I should be able to run all scripts (praat), as well as knit your reproducible report. I will try to do this, so test everything to make sure it works!
Evaluation
This is programming assignment 4 of 4. It is worth 10 of the 40 possible points. In order to receive full credit you must segment the audio files, extract the VOT data, load it into R, and complete a minimal analysis (descriptive stats, plots).
| Task | Points | Points earned |
|---|---|---|
Fork the pa_4 repo |
0.5 | |
| Correctly segment items | 2 | |
| Fix scripts and extract data | 2 | |
| Generate a hypothesis | 0.5 | |
| Calculate descriptives for VOT, present them in a table | 1 | |
| Plot VOT | 1 | |
| Include figure(s) of the acoustic data | 0.5 | |
| Evaluate your hypothesis | 0.5 | |
| Write a reproducible report | 1 | |
| Knit the reproducible report | 1 | |
| Total | 10 |
Tips
- Remember that Praat and R are picky regarding special characters and capitalized characters. Before you contact me (or others) for help, make sure you have reviewed previous programming assignments, the praat help manual, google, etc.
- You will not be able to finish this project the day it is due. The skills required to complete this assignment are cumulative, thus it will take a bit more time. Start early and think about the problems it poses.
- Regarding your hypothesis: do not go crazy and over think this. It can be a simple, obvious hypothesis, or something more risky/specific. Just remember that you have to be able to evaluate it (simple is usually better). If you know how to do t-tests, ANOVA’s, mixed effect models, etc., great… if not, also great. Evaluate your hypothesis based on your current skill set (descriptive stats and plots are fine!). Keep in mind what we have learned about stops in class and in the readings.
- Important tips and extra info here