INDG 401 Lecture Notes - Lecture 5: Grammaticality, Topicalization, Type I And Type Ii Errors
9 Jun 2018
School
Department
Course
Professor
Ling 450/650 Laboratory
Linguistics: Introduction
For software: mcgill.ca/software
- We will need Matlab, McGill gives free license to use it
- Optional to install
We will be using an Automatic aligner
- Montreal Forced Aligner
- Optional but recommended
Also: Stats software R
- Required
- Scripted program
● Quizzes and write-up must be done individually
○ At least draw on two research articles!
● On average, you need to commit 2-3 hours/week out of class for this course
● You can suggest your experiment
○ Traditional North American population
○ English/French maybe
● Grammaticality Judgements, prosodic prominence
● Prosody studies address, usually, both P and S sides
● Can use corpus data
Why Experiments?
Link:
http://prosodylab.cs.mcgill.ca/~chael/McGillLing450LaboratoryLinguistics2018/Lectures/1_Int
roduction/introduction.html
- Scope judgements: Somebody has read every paper - inverse scope, topicalization
- Quantifier
- Felicity judgement: Only everyone solved that problem - deliberate pragmatic effect
If fact is obvious, don’t need to run experiments
- Can draw generalizations even though exceptions - must apply to most of the
population
- John like movies
- Contextualize statements so all the participants are on the same page
Methodological pitfalls when analyzing experiments
Document Summary
We will need matlab, mcgill gives free license to use it. Quizzes and write-up must be done individually. At least draw on two research articles! On average, you need to commit 2-3 hours/week out of class for this course. Prosody studies address, usually, both p and s sides. Scope judgements: somebody has read every paper - inverse scope, topicalization. Felicity judgement: only everyone solved that problem - deliberate pragmatic effect. If fact is obvious, don"t need to run experiments. Can draw generalizations even though exceptions - must apply to most of the population. Contextualize statements so all the participants are on the same page. Effect size - gives us an estimation of what difference in height is there between men and women. Random variation - cluster around the diagonal line, but if it"s random it"ll vary in both directions. More below the line - replication yields lower effect size.
