Large-scale study of speech acts' development using automatic labelling

Abstract

Studies of children's language use in the wild (e.g., in the context of child-caregiver social interaction) have been slowed by the time- and resource- consuming task of hand annotating utterances for communicative intents/speech acts. Existing studies have typically focused on investigating rather small samples of children, raising the question of how their findings generalize both to larger and more representative populations and to a richer set of interaction contexts. Here we propose a simple automatic model for speech act labeling in early childhood based on the INCA-A coding scheme (Ninio et al., 1994). After validating the model against ground truth labels, we automatically annotated the entire English-language data from the CHILDES corpus. The major theoretical result was that earlier findings generalize quite well at a large scale. Our model will be shared with the community so that researchers can use it with their data to investigate various questions related to language use development.

A systematic review and Bayesian meta‐analysis of the development of turn taking in adult–child vocal interactions

“…Most of the studies included in the meta‐analysis presented several issues for making strong inferences: (i) they were cross‐sectional, or covered only short longitudinal timespans, (ii) they presented roughly aggregated data (e.g., age instead of measures of linguistic and social development); (iii) they provided group‐level summary statistics instead of individual‐level data; (iv) they contributed interaction‐level estimates, instead of turn‐by turn response latencies; and (v) the data lacked variability: 51 of the 78 effect sizes reported concerned US English (see also Figure for a representation of the geographical reach of the data). In order to advance our understanding of the development of turn‐taking, we need access to comprehensive cross‐linguistic and more representative (e.g., of socioeconomic, cultural, and ethnic diversity) datasets that provide turn‐by‐turn response latencies and the possibility to code for conversational moves (Bergey et al, 2021; Nikolaus et al, 2021), in combination with longitudinal assessments of linguistic and social development (Fusaroli et al, 2019; Naigles & Fein, 2017). A recent paper and R package has also been produced to provide streamlined standard ways of preprocessing the data (Casillas & Scaff, 2021), and other approaches are being developed to identify diverse types of interactional sequences and provide a partially automated coding of them (Bergey et al, 2021; Nikolaus et al, 2021).…”

“…In order to advance our understanding of the development of turn‐taking, we need access to comprehensive cross‐linguistic and more representative (e.g., of socioeconomic, cultural, and ethnic diversity) datasets that provide turn‐by‐turn response latencies and the possibility to code for conversational moves (Bergey et al, 2021; Nikolaus et al, 2021), in combination with longitudinal assessments of linguistic and social development (Fusaroli et al, 2019; Naigles & Fein, 2017). A recent paper and R package has also been produced to provide streamlined standard ways of preprocessing the data (Casillas & Scaff, 2021), and other approaches are being developed to identify diverse types of interactional sequences and provide a partially automated coding of them (Bergey et al, 2021; Nikolaus et al, 2021).…”

A systematic review and Bayesian meta-analysis of the development of turn taking in adult-child vocal interactions

“…While there are obvious data protection issues to be considered when sharing adult-child interactional data, there are several promising attempts (MacWhinney, 2000). A recent paper and R package has also been produced to provide streamlined standard ways of pre-processing the data (Casillas & Scaff, 2021), and other approaches are being developed to identify diverse types of interactional sequences and provide a partially automated coding of them (Bergey et al, 2021;Nikolaus et al, 2021).…”

Using video calls to study children’s conversational development: The case of backchannel signaling

“…Thus, we settled on a manageable sample size. That said, progress in automatic annotation of children's data (Sagae et al, 2007;Nikolaus et al, 2021;Long et al, 2022;Erel et al, 2022) should alleviate the constraint on large-scale data collection in future research. The current work is also supposed to contribute to this effort by providing substantial hand-annotated data that can be used for automatic model training and validation.…”