R and mixed models are becoming the de facto standard way of dealing with linguistic data, especially in fields such as phonetics. This two-day applied workshop will introduce participants to the linear model framework and its extensions, including generalized linear models and linear mixed effects models. Particular attention will be paid to the specificity of the analysis of phonetic (acoustic and articulatory) data.

Participants should bring their own laptop with R and RStudio installed. Package installation instructions will be distributed closer to the date of the workshop.

Location : ILPGA, salle Brunot, 19 rue des Bernardins, 75005 Paris

Day I – Monday, May 20th

9:30—12:00

Introduction to linear modeling with R

(tidyverse, R Markdown, multiple regression, categorical predictors, contrast coding schemes, centering & standardizing covariates, collinearity)

14:00—17:30

Linear models II + GLM intro

(inferential stats with regression, interactions, logistic regression, Poisson regression)

Day II – Tuesday, May 21th

9:30—12:00

Mixed model intro

(the independence assumption, fixed vs. random effects, random intercepts vs. random slopes, intercept/slope correlations, lme4 syntax)

14:00—17:30

Mixed models II

(afex package, mixed logistic regression exercise, dealing with convergence issues, rationale & philosophy for making decisions about what model to fit)