Competition between rhythmic and linguistic meaning revisited: the effect of task demands

ABSTRACT

This paper revisits the conclusion of our previous work regarding the dominance of meaning in the competition between rhythmic parsing and linguistic parsing. We played five-note rhythm patterns in which each sound is a spoken word of a five-word sentence. We asked listeners to indicate the starting point of the rhythm while disregarding which word would normally be heard as the first word of the sentence. In four studies, we varied task demands by introducing differences in rhythm complexity, rhythm ambiguity, rhythm pairing, and semantic coherence. We found that task complexity affects the dominance of meaning. We therefore amend our previous conclusion: when processing resources are taxed, listeners do not always primarily attend to meaning; instead, they primarily attend to the aspect of the pattern (rhythm or meaning) that is more salient.

KEYWORDS:

• Auditory necklaces
• Complexity
• Rhythmic organization
• Sentence processing
• Stroop effect

Notes

¹Although it is true that participants chose the necklace clasp more in the syncopated than isochronous condition in Study 1, Table 3 shows that overall listeners selected the necklace clasp more in the isochronous condition. This is especially noteworthy since the necklace clasp in the isochronous conditions was often the modal clasp, whereas it was the only relevant clasp for the syncopated conditions. It is unclear why the pattern was reversed in Study 1, but overall the pattern aligns with our conclusion that processing resources are more taxed in the syncopated condition than the isochronous condition.

²Due to a programming error, we could not to include the effect of syntax accuracy (with levels correct and missing article) on clasp choice as either a fixed effect or random effect. The effect of syntax accuracy thus remains an open question.

³Here and in Study 4, the Intercept effect represents the log-odds of choosing sentence clasp in the isochronous condition. In Study 2, the random condition is coded as the Intercept. In Study 3, the ambiguous condition is coded as the Intercept.

⁴The AN Type effect represents the difference in log-odds of choosing sentence clasp in the isochronous condition and syncopated conditions.

⁵We repeated this study with a between-participants design and found similar results. We do not report them because our small sample sizes led to large confidence intervals.

Additional information

Funding

This research was supported by the NSF Graduate Research Fellowship Program under Grant No. DGE-00809128 (LG); NSF Grant No. BCS 1027120 to the University of Virginia (PI: MK).