MetaLab

Interactive tools for community-augmented meta-analysis,
power analysis, and experimental planning in language acquisition research

13

Meta-analyses

252

Papers

887

Effect sizes

11,363

Participants

Meta-analyses currently in MetaLab:

Infant directed speech preference

Looking times as a function of whether infant-directed vs. adult-directed speech is presented as stimulation.

Label advantage in concept learning

Infants' categorization judgments in the presence and absence of labels.

Gaze following

Gaze following using standard multi-alternative forced-choice paradigms.

Vowel discrimination (native)

Discrimination of native-language vowels, including results from a variety of methods.

Vowel discrimination (non-native)

Discrimination of non-native vowels, including results from a variety of methods.

Phonotactic learning

Infants' ability to learn phonotactic generalizations from a short exposure.

Statistical sound category learning

Infants' ability to learn sound categories from their acoustic distribution.

Word segmentation

Recognition of familiarized words from running, natural speech using behavioral methods.

Online word recognition

Online word recognition of familiar words using two-alternative forced choice preferential looking.

Mutual exclusivity

Bias to assume that a novel word refers to a novel object in forced-choice paradigms.

Pointing and vocabulary (longitudinal)

Longitudinal correlations between pointing and later vocabulary.

Pointing and vocabulary (concurrent)

Concurrent correlations between pointing and vocabulary.

Sound symbolism

Bias to assume a non-arbitrary relationship between form and meaning ("bouba-kiki effect") in forced-choice paradigms.

Download data

Scatter plot of effect sizes over age

Funnel plot of bias in effect sizes

Violin plot of effect size density

Forest plot of effect sizes and meta-analysis model estimates

Meta-analytic model summary

Experiment planning

Select a meta-analysis and a set of moderators to see statistical power estimates using the estimated effect size (random effects) for that phenomenon.

Power plot of N necessary to achieve p < .05

Statistical power to detect a difference between conditions at p < .05. Dashed line shows 80% power, dotted line shows necessary sample size to achieve that level of power.

Experiment simulation

Run a simulation of a looking-time experiment, choosing an effect size and a number of participants per group. See the results of statistical comparisons for within-subjects effects (t-test) and for comparison with a negative control group (ANOVA interaction).


Simulated data

Simulated statistical tests

Meet the MetaLab team


Christina Bergman

École normale supérieure
chbergma@gmail.com
word segmentation
reproducibility
p-curves

Mika Braginsky

Stanford University
mika.br@gmail.com
infrastructure

Alejandrina Cristia

École normale supérieure
alecristia@gmail.com
IDS preference

Michael C. Frank

Stanford University
mcfrank@stanford.edu
developmental curves
p-curves

Molly Lewis

Stanford University
mll@stanford.edu
mutual exclusivity
label advantage
reproducibility

Page Piccinini

École normale supérieure
page.piccinini@gmail.com
developmental curves

Sho Tsuji

École normale supérieure
tsujish@gmail.com
phonemic discrimination