Pre-registration

[jvcasillas]

• [x] Create template
• [x] Determine modeling strategy
• [x] Power analysis
• [x] Outline
• [x] Write pre-reg

[kparrish92]

@jvcasillas added the pre-registration template in an Rmarkdown file and pushed to the repo

[jvcasillas]

Excellent. Moved to docs directory via https://github.com/RAP-group/empathy_intonation_perc/commit/d02a1727bc7c539c6c67162167d23789e3093690

[jvcasillas]

Outcome variables

• Response accuracy
• Response time

Predictors

• proficiency (lexTALE)
• empathy score (EQ)
• frase structure
  • narrow focus statements
  • broad focus statements
  • absolute interrogatives (yes-no questions)
  • partial interrogatives (wh-questions)

Analysis questions to consider for pre-reg:

• Separate models for each frase structure? (do we care about interactions between them?)
• Power analysis for two continuous variables (r-family es)
• Likelihood distributions for outcome variables (probably Bernoulli and shifted log-normal)
• What are weakly informative priors for these models?
• What is a reasonable ROPE for proficiency by empathy interaction?

[jvcasillas]

https://github.com/RAP-group/empathy_intonation_perc/blob/main/docs/pre_reg/preregistration_empathy.Rmd

@kag346 @isabelleschang

[laurafdeza]

There are two things I would change in the pre-reg:

1. Lines 106-107: maybe say we adapted Brandl et al. stimuli instead of we are using the same ones
2. Line 115: It reads as if all participants were native Spanish speakers. I would say "Spanish second language speakers"

[juanjgarridop]

Thank you @kag346 and @isabelleschang for completing the pre-registration!

Here are my comments/questions/suggestions regarding the pre registration:

1. Hypotheses (lines 31-38): The hypotheses mention sensitivity, accuracy, and reaction times. Are we measuring sensitivity by analyzing accuracy and rts? if that's the case, then I'd only put sensitivity in the hypotheses. Accuracy and reaction times can be spelled out in the predictions. In addition, the hypotheses mention phrases but not questions, and rising intonation but not falling intonation. I would modify the hypotheses this way:

H1: L2 learners of Spanish and Spanish monolinguals with a higher empathy quotient are more sensitive to intonation differences in Spanish statements and questions. (OR Higher empathy quotient leads to higher sensitivity to intonation differences in Spanish statements and questions in L2 learners of Spanish and Spanish monolinguals)

H2: L2 learners of Spanish and Spanish monolinguals with a higher level of proficiency in Spanish are more sensitive to intonation differences in Spanish statements and questions. (OR Higher proficiency level in Spanish leads to higher sensitivity to intonation differences in Spanish statements and questions in L2 learners of Spanish and Spanish monolinguals)

2. Line 48: As Nicole commented, is meta-analysis accurate for this study?

3. Study design (line 58): As Nicole commented, should we be more specific in this section and give all the details of the procedure and materials? If so, we can divide this part among some of us to write short paragraphs describing each of the screening and experimental tasks.

4. Explanation of existing data (line 103): we are not using any data that already exists, right? We are using their stimuli but not their data, right?

5. Line 115: As Laura commented, should we be more specific and mention a specific L1 and maybe country of residence? "adult English learners of Spanish living in the US" (or in an English-speaking country) and Spanish monolinguals (is there any criteria for the monolinguals?).

These are just suggestions.

[kparrish92]

Here are my comments! I mostly agree with the rest, but I have a few things that may have not been mentioned:

1. Line 17 - Title

The title should probably be a little more specific. Consider the title we used in the github homepage: Title: Production, perception, and processing of prosody in second language acquisition

2. Line 19 - Description

What is there looks good, but we could probably consider adding more detail to this section (groups, tasks and brief motivation)

3. Line 33 - H1

Similar to Juanjo: We could consider "Higher empathy L2 learners make use of intonational cues in question/statement disambiguation at lower levels of proficiency than lower empathy L2 learners".

4. Line 36 - H2

Agree with Juanjo - "L2 learners of Spanish and Spanish monolinguals with a higher level of proficiency in Spanish are more sensitive to intonation differences in Spanish statements and questions. (OR Higher proficiency level in Spanish leads to higher sensitivity to intonation differences in Spanish statements and questions in L2 learners of Spanish and Spanish monolinguals)"

5. Line 62 - Study Design

Agree with Nicole. We should consider adding more detail to this section by explaining each task in detail, what each task's purpose is, and making more clear the order of the tasks.

6. Line 103 - Explanation of existing Data

We could include this information in the study design section, since it isn't their data that we will use, but rather their stimuli as a basis for our own. Also, I agree with Nicole that we should point out that we have re-recorded the stimuli with new speakers and (we may) have made adjustments to them in content.

7. Line 123 - Sample size, Line 129 Sample size rationale

If we will do a power analysis, how will we determine effect size? The Brandl et al. study used means with pretty considerable variance (at least to my untrained eye), so I am unsure this would be a good place to render an effect size to use in our power analysis. What are the alternatives?