The Scope of Audience Design in Child-Directed-Speech: Parents Tailor Word Length for Adults versus Children

• In Event: 3-193 - The Moment-to-Moment Dynamics of Infant-Directed Speech: What Infants Hear and How They Learn

Integrative Statement

Humans share information efficiently and dynamically during natural communication, and a wide range of factors affect their success in doing so. For example, when conveying messages to others, we reduce and lengthen words differently depending on their predictability, frequency, and discourse status (Aylett & Turk, 2004; Fowler & Housum, 1987; Lam & Watson, 2010). Speakers may reduce because of automatic constraints on the production system (Kraljic, & Brennan, 2005) or because they tailor utterances for their listeners (Jaeger, 2010). Here, we examined parents’ word use across utterances in adult-directed speech (ADS) and child-directed speech (CDS) to determine which account is correct. If speakers reduce repeated words because of intrinsic features of their production systems, we should see reduction regardless of audience. If, however, speakers reduce in order to facilitate comprehension by their audience, we should see differences in reduction for adult vs. child listeners.

In two experiments, parents read instructions to an experimenter (adult-directed) and then to their own 2-3-year-old children (child-directed). In Experiment 1, parents (N=24) produced 32 sentences containing high-frequency target nouns. ‘Repeated’ trials were those in which successive utterances contained the same targets, e.g., “Get the cow. Can you put it under the big log? Now, put the cow in the purple castle.” ‘New’ trials were those in which two successive utterances contained different targets. Experiment 2 was identical except that parents (N=24) produced sentences containing lower-frequency nouns (e.g., brooch, felt).

We constructed mixed-effects models for each experiment to analyze log-transformed word durations. Fixed effects included audience (CDS, ADS), discourse status (repeated, new), and utterance (first, second). In Experiment 1, we found that parents reduced repeated information in both ADS (t=7.20, p<.001) and CDS (t=5.36, p<.001). But critically, parents reduced significantly less in CDS than in ADS (t=3.29, p=.005). The pattern of results was identical in Experiment 2 (Figure 1): Parents reduced repeated words regardless of audience, but they reduced less when talking to children. An additional finding in Experiment 2 was that words were longer in CDS than in ADS (t=5.90, p<.001).

Our results provide partial support for Fisher and Tokura’s (1995) claim that speakers have broad control over speech rate but do not have active control over finer-grained processes, evidenced by parents’ reduction of repeated words when speaking to young children. CDS is typically described as a clear instance of audience design that benefits young children’s learning (Liu et al., 2003; Schwab & Lew-Williams, 2016), yet parents reduced repeated words regardless of their child or adult audience. This is likely driven by an inherent automaticity in speakers’ production systems (Arnold & Watson, 2015; Kahn & Arnold, 2012). Critically, however, parents did engage in two distinct aspects of audience design: (1) they reduced referents less in CDS than in ADS across both experiments, and (2) in Experiment 2, they lengthened words in CDS, demonstrating recognition that children may need additional support to understand lower-frequency nouns. In summary, we found that parents’ skillful tailoring of speech for immature listeners is nested within automatic production processes.

Authors

• Nicholas Tippenhauer, Vanderbilt University
  Presenting Author

• Eva Fourakis, Princeton University
  Non-Presenting Author

• Casey Lew-Williams, Princeton University
  Non-Presenting Author

• Duane G. Watson, Vanderbilt University
  Non-Presenting Author