Science1 publisher2 min readPublished
Parent reports place the start of overheard word learning between 18 and 24 months
In 79 US toddlers surveyed three times, words that parents said their child mostly overheard from adult conversation were recognised at 24 and 30 months but not at 18. Every measure came from the parents.
The Scientist · Science desk
What happened
- Researchers at Chicago, The New School and Princeton surveyed the parents of 79 US toddlers at 18, 24 and 30 months, asking which household objects the child recognised and which words the child mainly overheard.
- Children at 24 and 30 months recognised words their parents classed as mostly overheard from adult talk, including pot, salad and charger, while the 18-month-olds did not.
- The team expected success at 18 months because lab experiments had already shown that 18-month-olds can pick up novel words from an exchange they were not part of.
Compiled by The ScientistSomething wrong?How this is made
Why it matters
- constraint Coaching or scoring caregiver input by how much adults say to the child is working inside less than half of what the child hears, if the more-than-half exposure figure holds.
- contradiction Lab experiments credit 18-month-olds with learning from overheard speech, and this home-based measure did not detect it until 24 months, so lab and home give different onset ages.
- decision A causal claim about overhearing now needs a design that varies exposure; parent report of both the word type and the recognition shows only that the two move together.
Lab work on overhearing labels the target word clearly and keeps the object visible [13]. Home is messier. The author team told the Society for Research in Child Development [19] that one likely explanation for the 18-month result is that "natural overheard speech presents a tougher challenge than the overheard speech researchers create in experiments" [11], and that it "doesn't come with the exaggerated, sing-song tone adults use when talking to babies, for instance" [12].
Parents supplied both halves of the comparison. They reported which household words their child mainly overheard from adult talk, such as pot, salad and charger, and which were mainly said to the child, such as zipper, mirror and oatmeal [3]. They also reported whether their child recognised each object [4]. Both the word types and the recognition come from one informant [18].
Order of events matters for the second finding. Among the 18-month-olds, the ones who had larger vocabularies by 24 months were also the ones with better recognition of overheard words and a narrower gap between the two word types [6]. The vocabulary measure that sorts them comes six months after the recognition measure it is being matched against [21]. Two views of the same developing ability move together. The design leaves open whether overhearing built the vocabulary or the vocabulary opened up overhearing [6].
The premise in the first line of the summary is what gives the result its reach: overhearable speech, meaning talk between caregivers in the child's presence, is more than half of a child's language exposure [1]. If that holds, speech addressed to the child is under half of what reaches its ears [16]. The authors' reading is that adult conversation carries words that never come up in talk to a one-year-old, "wallet" and "receipt" among them [15]. "They first learn from the speech we offer them directly; then, somewhere around their second birthday, they start helping themselves to the conversations we never meant for them to hear," the team said [14]. The interview gives the direction of each finding; the authors did not give effect sizes [20].
The paper is in Child Development, by Yuchen Jin and Marisa Casillas of the University of Chicago, Ruthe Foushee of The New School for Social Research and Kennedy Casey of Princeton [8]. The window it dates, the six months between the 18-month and 24-month visits [17], comes from 79 predominantly White and mixed-race children learning English in the United States [4].
What to watch
- Whether exposure measured from daylong home audio recordings, instead of parent report, also puts the onset after 18 months.
- Whether the pattern holds in communities where adults address young children less often, and in languages other than English.
- Whether the 24-month vocabulary moderator survives in a sample larger than 79 children.