Day by day, hour by hour: Naturalistic language input to infants

Abstract

Measurements of infants' quotidian experiences provide critical information about early development. However, the role of sampling methods in providing these measurements is rarely examined. Here we directly compare language input from hour-long video-recordings and daylong audio-recordings within the same group of 44 infants at 6 and 7 months. We compared 12 measures of language quantity and lexical diversity, talker variability, utterance-type, and object presence, finding moderate correlations across recording-types. However, video-recordings generally featured far denser noun input across these measures compared to the daylong audio-recordings, more akin to 'peak' audio hours (though not as high in talkers and word-types). Although audio-recordings captured ~10 times more awake-time than videos, the noun input in them was only 2-4 times greater. Notably, whether we compared videos to daylong audio-recordings or peak audio times, videos featured relatively fewer declaratives and more questions; furthermore, the most common video-recorded nouns were less consistent across families than the top audio-recording nouns were. Thus, hour-long videos and daylong audio-recordings revealed fairly divergent pictures of the language infants hear and learn from in their daily lives. We suggest that short video-recordings provide a dense and somewhat different sample of infants' language experiences, rather than a typical one, and should be used cautiously for extrapolation about common words, talkers, utterance-types, and contexts at larger timescales. If theories of language development are to be held accountable to 'facts on the ground' from observational data, greater care is needed to unpack the ramifications of sampling methods of early language input.

Figure 1.

Noun count measures across audio-recordings and videos. Top row depicts daylong audio data; bottom row shows the 3 hour-long annotations: “same” and “top” are the two peak audio times, and “video” indicates the video data. Upper panel labels indicate annotated sample length (day or hour); the bottom panel labels reflects measure type (op = object presence; utt = utterance-type, quant = quantity, Nspeakers = number of speakers). Bars (left to right) appear in legend order (top to bottom) in both color (count measures) and opacity (time sample: day, top-hour, same-hour, or video).

Figure 2.

Noun count measures normalized by recording length, for audio-recordings (solid borders) and videos (dashed borders). Normalized counts were calculated by dividing raw counts (see Fig 1.) by non-silent recording minutes. op = object presence; utt = utterance-type, quant = quantity, Nspeakers = number of speakers. Bars (left to right) appear in legend order (top to bottom). All measures differed significantly across recording-types except nouns from fathers.

Figure 3.

Normalized count correlations between audio- vs. video-recordings. Each point indicates nouns per minute of recording for each child, averaged across months 6 and 7, for each measure. Point-shape indicates measure type. Robust linear correlations are plotted for visualization only; non-parametric correlations (Kendall) were computed for analysis, showing that all correlations were significant except nouns from fathers and in singing.

Figure 4.

Utterance-type proportions across audio-recordings (daylong, “same” hour and “top” hour) and videos (indicated by line-type). Utterance-types are in legend order top to bottom. Videos contained a significantly more questions and fewer declaratives than the audio-recording time samples.

Figure 5.

Log-scaled counts of the top 100 words in audio- and video-recordings. Each node represents the averaged count, across all participants in both months, of each noun (0.1 was added before taking logs to include 0 counts.) Words in blue occurred 0 times in one recording type; words in pink were attested in both recording types. Nodes are jittered for visual clarity, with grey lines indicating node location on axes.

Figure 6.

Correlations of the frequencies of the top 100 words in audio- vs. video-recordings. Each node represents one word averaged across all participants in both months.

Figure 7.

Top 10 words by recording type and time sample. Each node represents the frequency count of each top audio or video word over both months (x-axis) and the number of families where that word was said (out of 44) across months (y-axis).