Using Prosody to Predict Upcoming Referents in the L1 and the L2

Abstract

While monolingual speakers can use contrastive pitch accents to predict upcoming referents, bilingual speakers do not always use this cue predictively in their L2. The current study examines the role of recent exposure for predictive processing in native German (L1) second language learners of English (L2). In Experiment 1, participants followed instructions to click on two successive objects, for example, Click on the red carrot/duck. Click on the green/GREEN carrot (where CAPS indicate a contrastive L + H* accent). Participants predicted a repeated noun following a L + H* accent in the L1, but not in the L2, where processing was delayed. Experiment 2 shows that after an exposure period with highly consistent prosodic cues, bilinguals engaged in predictive processing in both their L1 and L2. However, inconsistent prosodic cues showed different effects on bilinguals’ L1 and L2 predictive processing. The results are discussed in terms of exposure-based and resource-deficit models of processing.

Adaptation in Predictive Prosodic Processing in Bilinguals

“…In contrast, the available evidence suggests that L2 listeners do not always use contrastive pitch accents to predict upcoming referents (Klassen, 2015;Namjoshi, 2015;Takeda, 2018;Perdomo and Kaan, 2019;Foltz, 2020;Nakamura et al, 2020). Most relevantly for the current study, Foltz (2020) tested German-English bilingual listeners' processing of contrastive pitch accents in both of their languages. She found that participants used L+H * accents to predict upcoming referents in their L1, but not consistently in their L2, even though both languages mark contrastiveness through the same L+H * cue.…”

“…She found that participants used L+H * accents to predict upcoming referents in their L1, but not consistently in their L2, even though both languages mark contrastiveness through the same L+H * cue. Specifically, Foltz (2020) found that L2 listeners initially showed no evidence for predictive processing, but for facilitative processing, where the pattern expected for predictive processing occurred only after identifying segmental information from the final noun had come in. Specifically, when hearing Click on the red banana.…”

“…Klick die rote… (literally: Click the purple scissors on. Click the red…; Weber et al, 2006a;Foltz, 2020). In both cases, the predictive L+H * cue occurred on the word immediately preceding the predicted noun, suggesting that native listeners can use the contrastive pitch accent cue rapidly to predict upcoming referents in the discourse.…”

“…While a large body of evidence suggests that L2 speakers engage in predictive processing to a lesser extent compared to native speakers (Kaan, 2014;Grüter et al, 2017), we so far have relatively little knowledge of the kinds of processing situations in which L2 learners would or would not engage in predictive processing and why this would be the case. There is also evidence that L2 learners adapt to the input differently in their L1 compared to their L2 (Foltz, 2020). One possibility is that L2 learners have difficulties in generating a prediction error when faced with unpredicted input, such that they adjust their predictions less than native speakers or not at all in the face of a prediction error.…”

“…The current study uses a Bayesian adaptation model (Delaney-Busch et al, 2019;Ness and Meltzer-Asscher, 2021a), which includes trial-by-trial adaptation to the input and incorporates prediction error, to explore if such a model is compatible with the processing of prosodic cues observed in both the L1 and the L2. Specifically, the study uses the eye-tracking data from German-English bilinguals from Foltz (2020) to explore the hypotheses that a strong prediction error leads to delayed processing of the unpredicted input and causes adaptation in the forms of reduced prediction in future trials, such that the number of disconfirmed predictions affects processing.…”