What can we learn about two million priming values? An update to the Semantic Priming Across Many Languages Project (SPAML)

Erin M. Buchanan & The Psychological Science Accelerator

Harrisburg University

Many Thanks

• The Psychological Science Accelerator
• ZPID: https://leibniz-psychology.org/
• Harrisburg University
• Over 100 labs and 300+ collaborators
• Many translators
• The admin team

Semantic Priming

FASTER

SLOWER

SPAML Stimuli

• Selected by language word frequency and similarity using Open Subtitles
• 1000 cue-target pairs
• Matched across 31 languages/dialects
• Required: 10 languages for registered report

SPAML Procedure

Participant Level Data

• 125 labs have received a data collection link
• 27121 total participants
• 23001 total “usable participants”
  • At least 18
  • At least 100 trials
  • At least 80% accuracy
• Unfortunately, this is lower data retention than we anticipated
  • SPP was approximately 95%
  • Ours is 84%
• Total number of trials: 17091940: 15953628 usable

Participant Level Data

• Also Thai, Hindi, Slovenian, Finnish

Participant Level Data

• For this talk, we are going to split the data into:
  • Native: Took the study in their listed native language, browser matched language of study
  • Non-Native/Multiple: Took the study in a different language or listed multiple native languages or included a mismatch of browser and language of study

Participant Level Data

Participant Level Data

Priming Calculation

• By item:
  • Matched on the target and then averaged
  • {SPOON}-DOG trial for DOG minus {CAT}-DOG trial for DOG
• By condition:
  • Overall unrelated condition mean on targets compared to related condition mean on targets
• For this presentation, required n >= 20 for target word

Priming Level Data: By Item

Priming Level Data: By Condition

Priming Level Data: By Condition

Priming Level Data

• By Item:
  • Priming ~ .11
  • No differences in language groups
• By Condition:
  • Priming ~ .11
  • No difference/interaction in language groups

Priming Level Data

Priming Level Data: Items Matched

Priming Level Data: Items Matched

Cue	Target
washington	boston
dude	bro
dog	cat
november	december
factor	element
internal	external
near	far
male	female
two	four
kilo	gram
sandwich	hamburger
cocaine	heroin
romeo	juliet
june	july
seminar	lecture
advertising	marketing
ketchup	mayonnaise
second	minute
september	october
garlic	onion
property	owner
vagina	penis
increase	reduce
pub	restaurant
ninja	samurai
stand	sit
daughter	son
saturday	sunday
what	that
apply	use
sherlock	watson
red	yellow
today	yesterday

Final Thoughts

• We have a lot of data! It is so exciting!
• There is priming!
• It appears to be consistent across native and multi-lingual people
• Differences in amount by language

Questions?

• Thank you for listening!
• All PSA collaborators are listed with their author information online