Positional biases in predictive processing of intonation

ABSTRACT

Real-time speech comprehension is challenging because communicatively relevant information is distributed throughout the entire utterance. In five mouse tracking experiments on German and American English, we probe if listeners, in principle, use non-local, early intonational information to anticipate upcoming referents. Listeners had to select a speaker-intended referent with their mouse guided by intonational cues, allowing them to anticipate their decision by moving their hand toward the referent prior to lexical disambiguation. While German listeners (Exps. 1–3) seemed to ignore early pitch cues, American English listeners (Exps. 4–5) were in principle able to use these early pitch cues to anticipate upcoming referents. However, many listeners showed no indication of doing so. These results suggest that there are important positional asymmetries in the way intonational information is integrated, with early information being paid less attention to than later cues in the utterance. Open data, scripts, and materials can be retrieved here: https://osf.io/xf8be/.

KEYWORDS:

• Prosody
• intonation
• sentence comprehension
• rational analysis
• mouse tracking‌

Acknowledgment

Timo Roettger’s work was supported by the “Zukunftskonzept” of the University of Cologne as part of the Excellence Initiative. Michael Franke’s work was supported by the Priority Program XPrag.de (DFG Schwerpunktprogramm 1727). We would like to thank Nastassja Bremer and Kim Rimland for their help during data collection, as well as three anonymous reviewers, and the editor for their insightful comments and suggestions. All remaining errors are our own. Author contribution according to CRediT: TBR: Conceptualisation, Methodology, Resources, Data Curation, Writing – Original Draft, Writing – Review & Editing, Visualisation, Supervision, Project administration, Funding acquisition. MF: Conceptualisation, Methodology, Writing – Review & Editing. JC: Conceptualisation, Resources, Writing – Review & Editing, Supervision, Funding acquisition.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 For our analyses, we further used the following R packages: ggbeeswarm (Clarke & Sherrill-Mix, 2017), ggpubr (Kassambara, 2020), readbulk (Kieslich & Henninger, 2016), rstan (Stan Developer Team, 2020), rstudioapi (Ushey et al., 2020), stringr (Wickham, 2019), and tidyverse (Wickham et al., 2019).

2 Here, “heading towards the target” is operationalized by approximating the first derivative to the x- and y-coordinates of a trajectory; see function “get_TTT_derivative()” in included analysis scripts.

3 Comparisons between the model predictions and raw data indicate that the data is unlikely to be normally distributed around a single predicted mean value. Rather, it appears as if multiple distinct processes are responsible for the generation of the data. Given the nature of our primary measurement, the turn-towards-the-target (TTT), there are several instances in which listeners randomly drift toward the correct response early on (and not turning back), resulting in very early TTTs that are generated by chance rather than a genuine anticipation of the referent based on acoustic information. Given that these random drifts will occur equally often across conditions, they will not confound the comparison of groups.

4 The switch to American English was partly driven by pragmatic constraints with the first author changing institutions from the University of Cologne to Northwestern University. However, given the similarity between the German and the English intonation system, we are convinced that our findings can be informative about how listeners integrate intonational information across different sentence positions. We would like to emphasize that there are many differences between these two languages and their respective phonological systems, thus we do not consider our data as reliable evidence for cross-linguistic differences in the processing of intonation between these languages.

5 One might consider the contour in the PRENUCLEAR condition to be marked by double focus: An accented subject noun contrasts the wuggy on the screen with other wuggies seen during the experiment, while the accented object marks the referent as contrastive with regard to the topic in the preceding question. Note that this interpretation does not disqualify our claim that the accent on the subject referent is an early cue to the object referent, even if at the same time it might mark contrastive focus on “one”.