It's a 144 word list with:
48 pseudo words
48 non-cognates
48 cognates (split between 24 English-Spanish/2-way and 24 English-Spanish-German/3-way)
The cognates are all 100% orthographic overlap, and I have not yet examined phonological overlap.
Pilot data of 17 L2 Spanish speakers (L1 American English) showed an effect for both 2-way and 3-way cognates (we would expect a similar effect for both of these conditions for this group, while the L3 group is predicted to show a greater effect for 3-way cognates than 2-way. Participants were only included if the got at least 50 total answers correct.
A frequentist model shows an effect for both cognate types (log reaction time ~ cognate type).
Here are the random effect for the stimuli by word type. Each plot text shows each words impact on the fixed effect.
Based on visual inspection, I don't think any of them seem to be overly impactful. (One taboo word will need to be replaced in the 3-way list..oops.)
Finally, I looked at how the size of the facilitation effect changed as the number of total correct answers increased (a proxy for proficiency for pilot purposes). The effect trends in the right direction for both cognate types, and there is an effect in a linear model when they are pooled.
From here, all that we need to do is correct the single taboo word in the list and incorporate any recommendations that you have, and I can get some L3 data! I was thinking that it would also be ideal to have L2 speakers of Spanish as well, if you have access to some at Furman.
P.s. A few notes on the previous pilot data (with a long word list and cognates that were > .7 in orthographic overlap). There was not a continuous effect of overlap, and the 100% overlap tokens appeared not to show a difference. Moving forward, we could consider using this list (maybe even on the same people) in addition to our current one, to show that there are certainly list composition effects.
Greetings! (edited Jun 8th) I've created a new word list that appears to be showing the cognate effect in L2 Spanish. It's available here: https://github.com/kparrish92/l3_cognate_study_pilot/blob/main/data/stimuli/word_list.csv.
It's a 144 word list with: 48 pseudo words 48 non-cognates 48 cognates (split between 24 English-Spanish/2-way and 24 English-Spanish-German/3-way)
The cognates are all 100% orthographic overlap, and I have not yet examined phonological overlap.
Pilot data of 17 L2 Spanish speakers (L1 American English) showed an effect for both 2-way and 3-way cognates (we would expect a similar effect for both of these conditions for this group, while the L3 group is predicted to show a greater effect for 3-way cognates than 2-way. Participants were only included if the got at least 50 total answers correct.
A frequentist model shows an effect for both cognate types (log reaction time ~ cognate type).
Here are the random effect for the stimuli by word type. Each plot text shows each words impact on the fixed effect. Based on visual inspection, I don't think any of them seem to be overly impactful. (One taboo word will need to be replaced in the 3-way list..oops.)
Finally, I looked at how the size of the facilitation effect changed as the number of total correct answers increased (a proxy for proficiency for pilot purposes). The effect trends in the right direction for both cognate types, and there is an effect in a linear model when they are pooled.
From here, all that we need to do is correct the single taboo word in the list and incorporate any recommendations that you have, and I can get some L3 data! I was thinking that it would also be ideal to have L2 speakers of Spanish as well, if you have access to some at Furman.
P.s. A few notes on the previous pilot data (with a long word list and cognates that were > .7 in orthographic overlap). There was not a continuous effect of overlap, and the 100% overlap tokens appeared not to show a difference. Moving forward, we could consider using this list (maybe even on the same people) in addition to our current one, to show that there are certainly list composition effects.