'Eye-tracking evidence for Turkish heritage speakers predictive use of case-marking in German and Turkish' (AsPredicted #85339)
Author(s) This pre-registration is currently anonymous to enable blind peer-review. It has 5 authors.
Pre-registered on 2022/01/17 - 07:34 AM (PT)
1) Have any data been collected for this study already? No, no data have been collected for this study yet.
2) What's the main question being asked or hypothesis being tested in this study? Özge et al. (2019) and Özge et al. (2021) found consistent effects that monolingual child and adult speakers of Turkish and German can use case-marking predictively to determine the thematic roles (agent vs theme) of given NPs (subject vs object) in sentences with non-canonical word orders. It is unclear whether these effects replicate with bilingual heritage speakers. We assume that heritage speakers are fully competent bilingual speakers who have fully acquired core grammatical features like case in both their languages.
Predictions:
1. Based on Özge et al. (2019), heritage speakers of Turkish will be able to use case-marking predictively to determine thematic roles in incremental sentence-processing of SOV and OSV sentences.
2. Based on Özge (2021), heritage speakers of Turkish in Germany will be able to use case-marking predictively in their majority German to determine thematic roles in incremental sentence-processing of SOV and OSV sentences.
3. Consequently, speaker group (monolingual German/Turkish vs. heritage Turkish German/Turkish) will not be a significant predictor for the ability of speakers to use case-marking to predict thematic roles.
3) Describe the key dependent variable(s) specifying how they will be measured. Behavioral measure to track attention/competence: Accuracy of correctly matching the action picture using yes (F key) and no (J key) responses after each trial to the sentence that they previously heard.
Fixational measure: Eye-gazes as Agent Preference coded as 1 for looks at the potential agent and 0 for looks at the potential patient (following Özge et al., 2019; Özge et al. 2021)
4) How many and which conditions will participants be assigned to? There are three manipulated factors with two levels each, respectively:
1. Word order in verb-final sentences manipulated by case-marking (SOV / NOM-ACC-V vs. OSV / ACC-NOM-V).
2. Language (German vs Turkish).
3. Group (heritage vs monolingual).
5) Specify exactly which analyses you will conduct to examine the main question/hypothesis. Accuracy of correctly matching the action picture will be calculated and no inferential analysis is planned for this variable.
Data processing and analyses will be conducted in R using at least tidyverse, lme4, lmerTest and eyetrackR packages. Fixation times will be divided into 100-ms time-windows and matched with the sequence of the stimuli sentences. Within each time-window, we'll calculate the measure for our dependent variable, Agent Preference, given the description in Özge et al. (2019)'s Section 2.2. Then, we'll analyze the data using generalized or additive regression models with (the interaction of) Case Marking on the first NP and Time Window of Interest (from the onset of the first NP to before the onset of of the second NP), as well as Group (heritage vs. monolingual), and Language (German vs. Turkish) as independent variables and random intercepts/slopes for Participant and Item.
6) Describe exactly how outliers will be defined and handled, and your precise rule(s) for excluding observations. Looks after the onset of the second NP will be excluded as they don't serve to test the predictions since they clearly identify the second NP. The same exclusion and reason follows for looks to the Topic or empty screen during trails.
Participant data will be excluded if the Accuracy of correctly matching the action picture is below 70 percent for a given participant.
7) How many observations will be collected or what will determine sample size? No need to justify decision, but be precise about exactly how the number will be determined. This study aims to replicate effects found in Özge et al. (2019) with N=21 to N=40 for Turkish and Özge et al. (2021) N=20 for German. Based on these replicated effects, we target approximately 30 participants per group (monolingual, heritage) and two languages (German, Turkish). In total, this yields 120 participants.
If we manage to have collected data from 20 participants in a given group, but not the targeted number of 30 by 1 July 2023, we'll stop data collection on this date for practical purposes.
8) Anything else you would like to pre-register? (e.g., secondary analyses, variables collected for exploratory purposes, unusual analyses planned?) N/A
