Poster #55 - Infant-Caregiver Vocal Coordination in Daily Interaction in Tseltal and Canadian English

• In Event: Poster Session 18

Abstract

Tseltal Mayan children receive less direct adult speech input than children in many Western communities, but do not show delays in the onset of major communicative benchmarks (Casillas, Brown, & Levinson, 2020). We investigate how children from this same small-scale, subsistence farming community in Southern Mexico engage in vocal coordination with their mothers. We compare their patterns of coordination to Canadian infant-mother dyads. Six Tseltal and six English (Low-SES samples) learning infants’ acoustic-prosodic coordination with their mothers was analyzed. Whereas infants occasionally aligned to their mothers in both groups, mothers only aligned to their infants in the Canadian sample.
Data come from the Casillas HomeBank Corpus (Casillas, Brown, & Levinson, 2017) and McDivitt HomeBank Corpus (McDivitt, & Soderstrom, 2016). Day-long audio recordings of six Tseltal (age range 0;2-1;10 years) and six Canadian-English learning infants (age range 0;2-1;07 years) were collected at their homes. The input for analysis includes utterance-level pitch maximum (f0_max), minimum (f0_min), mean (f0_mean), and range (f0_range).
We investigate vocal coordination on two levels: First, we measured alignment at speaker switches, which we define as instances where the pause between the two speakers’ turns is shorter than one second. We distinguish between mother-initiated (M-I) and infant-initiated (I-M) switches. Alignment was measured using the correlations of the prosodic features of the two speakers. Second, on the conversational block level, to measure whether the two parties dynamically affect each other’s prosodic features. Conversational blocks were defined as sequences containing at least two speaker switches with all pauses between shorter than 3 seconds. On this block level, we first compared the mean similarities (negative differences) of the prosodic features over each block across the two cultures and then assessed entrainment by correlating the block length (number of switches) and the prosodic similarity at each switch. Analyses were corrected for multiple comparisons using Benjamini/Hochberg correction within each group.
Results and discussion
At the switch-level, a weak but significant positive correlation was found in f0_max of Tseltal M-I switches. However, no significant results were found in Tseltal I-M switches. As for English dyads, significant correlations were found in all I-M switches, while only f0_range was significantly correlated in M-I switches (Table 1). On the conversational block level, the mean similarities of the four prosodic features from Tseltal dyads were significantly lower than those of English (p_range=.039, p_max=.048, p_min=.033, p_mean=.026, two-tailed T-test). However, a significant entrainment effect over blocks was found in Tseltal, but not in English (Fig.1). (Significance: p < .05 *; p < .01 **)
The results reveal that maternal pitch alignment occurs in English but not Tseltal, while infant alignment to mothers is weak but similarly present across cultures. Interestingly, Tseltal dyads were more likely than Canadian ones to show increases in pitch alignment over longer sequences of interaction. These findings, while based on a small sample, underline cultural differences in caregiver input. Further research on infants’ active role in the interaction may reveal mechanisms by which Tseltal, Canadian, and other infants robustly learn from a wide variety of experiences.

Authors

• Jiarui Li, The University of Tokyo
  Presenting Author

• Marisa Casillas, University of Chicago

• Melanie Soderstrom, University of Manitoba

• Yukie Nagai, The University of Tokyo

• Sho Tsuji, The University of Tokyo