We introduce a new resource: the SAYCam corpus. Infants aged 6-32 months wore a head-mounted camera for approximately 2 hours per week, over the course of approximately two and a half years. The result is a large, naturalistic, longitudinal dataset of infant- and child-perspective videos. Transcription efforts are underway, with over 200,000 words of naturalistic dialogue already transcribed. Similarly, the dataset is searchable using a number of criteria (e.g., age of participant, location, setting, objects present). The resulting dataset will be of broad use to psychologists, linguists, and computer scientists.

To help children navigate their social environments, adults must understand what children know about race, and when they acquire this knowledge. Across three pre-registered studies, we tested US adults’ knowledge of when children first use race to categorize and ascribe traits to others. Participants wildly—and uniquely—misjudged children’s abilities to process race. This inaccuracy was consequential: it was a stronger predictor of the preference to delay conversations about race with children than other factors previously theorized to underlie adults’ reluctance to talk about race. And, this relation was causal. Our data suggest that fundamental misunderstandings about children’s capacities to process race are pervasive in the US population and may delay when adults engage children in important conversations about race.

We test the hypothesis that children acquire the successor function — a foundational principle stating that every natural number n has a successor n+1 — by learning the productive linguistic rules that govern verbal counting. Previous studies report that speakers of languages with less complex count list morphology have greater counting and mathematical knowledge at earlier ages in comparison to speakers of more complex languages (e.g., Miller & Stigler, 1987). Here, we tested whether differences in count list transparency affected children’s acquisition of the successor function in three languages with relatively transparent count lists (Cantonese, Slovenian, and English) and two languages with relatively opaque count lists (Hindi and Gujarati). We measured 3.5- to 6.5-year-old children’s mastery of their count list’s recursive structure with two tasks assessing productive counting, which we then related to a measure of successor function knowledge. While the more opaque languages were associated with lower counting proficiency and successor function task performance in comparison to the more transparent languages, a unique within-language analytic approach revealed a robust relationship between measures of productive counting and successor knowledge in almost every language. We conclude that learning productive rules of counting is a critical step in acquiring knowledge of recursive successor function across languages, and that the timeline for this learning varies as a function of counting transparency.

By around the age of 5½, many children in the US judge that numbers never end, and that it is always possible to add +1 to a set. These same children also generally perform well when asked to label the quantity of a set after 1 object is added (e.g., judging that a set labeled “five” should now be “six”). These findings suggest that children have implicit knowledge of the “successor function”: every natural number, n, has a successor, n+1. Here, we explored how children discover this recursive function, and whether it might be related to discovering productive morphological rules that govern language-specific counting routines (e.g., the rules in English that represent base 10 structure). We tested 4- and 5-year-old children’s knowledge of counting with three tasks, which we then related to (1) children’s belief that 1 can always be added to any number (the successor function), and (2) their belief that numbers never end (infinity). Children who exhibited knowledge of a productive counting rule were significantly more likely to believe that numbers are infinite (i.e., there is no largest number), though such counting knowledge wasn’t directly linked to knowledge of the successor function, per se. Also, our findings suggest that children as young as four years of age are able to implement rules defined over their verbal count list to generate number words beyond their spontaneous counting range, an insight which may support reasoning over their acquired verbal count sequence to infer that numbers never end.

When children acquire language, they often learn words in the absence ofdirect instruction (e.g., “This is a ball!”) or even social cues toreference (e.g., eye gaze, pointing). However, there are few accounts ofhow children do this, especially in cases where the referent of a new wordis ambiguous. Across two experiments, we test whether preschoolers (2- to4-year-olds; *n* = 239) can learn new words by inferring the referent of anew word from the surrounding linguistic discourse. Across two experiments,we show that children as young as two can learn a new word from thelinguistic discourse in which it appears. This suggests that children usethe linguistic discourse in which a word appears – via discoursebootstrapping – to learn new words.