How Many Words Do Children Need?

At my town’s public library, one wall of the children’s section displays posters declaring, “Babies need words every day,” encouraging me to sing, talk, and read with my child. On the radio recently, I heard a PSA begging me to talk to my child as much as possible. You may have noticed similarly encouraging messages from your library, radio, pediatrician, and social media. If you’ve scratched the surface of the messages looking for answers about where they come from, you may have also heard of the “30 million word gap” which suggests that children living in poverty hear millions fewer words from their parents, impairing their language development compared to peers from wealthier backgrounds. The news coverage of this phenomenon can be controversial – blaming low-income parents for depriving their children of language — as well as overwhelming or confusing: How much talking is enough for my child?

The claims about the 30 million word gap originate from an article and book published in the 1990’s by Betty Hart and Todd R. Risley. The book, with the delightfully unassuming title of “Meaningful Differences in the Everyday Experience of Young American Children,” has been cited time again by psychologists and policy-makers to emphasize the importance of exposing children to a high volume of language early in life. The researchers were interested in improving language outcomes in low-income children in a preschool in Kansas, but found that the interventions they used in classrooms did not suffice. They became curious about what other influences the children had on their language, and designed a study to examine language in the home.

Hart and Risley collected data during the 1960’s and 70’s from 42 families, from the time a child was about 9 months of age until they turned 3 years old, with 13 of those families being categorized as “professional,” 23 classified as “working-class,” and 6 as “welfare.” Research assistants undertook the massive task of visiting families monthly, tape recording and writing down all language and interactions that occurred for an hour, and typing up the observations. (Given the technology at the time, a one-hour visit could take eight hours to transcribe.) In Hart and Risley’s book, they reported many results, including the often-cited differences in the sheer number of words that children from different participant groups heard. In the graph below, you can see that children with high socioeconomic status (SES) families heard the most words (cumulatively over time), followed by middle and then lower SES. If you look closely at the point when the children were 36 months old, you will see that the gap between the “high SES” and “low SES” lines is about 30 million – hence the number we’ve heard so much about.

Number of words addressed to children in the three SES groups by age

Source: Emma Kelty-Stephen, adapted from Hart & Risley, 1995

The next question is, clearly, does this matter? If a child simply hears a greater number of words over the early months of life, what difference will it make? Hart and Risley’s data suggest that there is, as their book title states, a “meaningful difference." In this graph, you can see that the number of words a child has in their vocabulary is increasingly different over time, depending on their SES group; the difference could mean that a high SES child knows twice as many words as a lower SES child. It is clear from Hart and Risley’s work that they were not blaming low SES parents for the lack of words addressed to children, and they make a point to mention loving, high-quality interactions between parents and children as unrelated to education and wealth level.

Number of words in children's vocabularies in the three SES groups by age

Source: Emma Kelty-Stephen, adapted from Hart & Risley, 1995

The sheer amount of time and effort that went into this study, especially given the technology of the time, is astounding. However, the work has recently been criticized for its small sample size (the low SES group had just 6 families in it) and for having other potential confounds (all low SES families were African-American). Another key criticism is that there is too much focus on the sheer number of words that children heard – isn’t there anything important about the kind of words they heard? Hart and Risley did think so: Their book contains analysis of parts of speech and whether parents used more encouraging or discouraging remarks to their children. Current researchers have been exploring ways of fine-tuning those analyses, developing measurements of language quality and using sophisticated statistical modeling to examine how predictive language input is to later child language ability.

Meredith Rowe of Harvard University published a study in 2015 in the journal Child Development tracking the roles of language quantity and quality in children’s vocabulary development. Quality in this case does not mean that language was considered “better” or “worse” for children, but rather that it contains certain elements, such as a word being rare or an utterance being decontextualized. Decontextualized utterances by parents add information beyond the here-and-now, challenging the child by referring to objects or events that aren’t in their direct experience, such as an imaginary scenario or something that happened in the past or may happen in the future. By tracking 50 parent-child pairs at three visits, from toddlerhood to preschool, Rowe found fascinating connections between the kind of language children heard and the number of words they knew at later ages.

Using similar techniques to Hart and Risley’s, Rowe found a wide variation in parent talk: Parents used as few as 400 words in 90 minutes and as many as 9000. Parents with more education (used as a way of getting at SES) produced more language and more decontextualized utterances, but the relationship did not seem consistent at each age. Rowe used statistical modeling to look at how parent language at each age related to their child’s vocabulary at later ages, controlling for parent education levels and the child’s initial vocabulary when finding predictors. This means that we can see how much quantity and quality matter in a child’s vocabulary above and beyond their parent’s education level, and how many words they knew in the first place. The careful analysis in the report indicates that parent education and sheer quantity of words to very young children (18 months old) are important indicators of child vocabulary later on, but that as a child gets older, the wider range of rare words and decontextualized speech becomes more important when predicting vocabulary at later ages. These findings give us a nuanced developmental picture of vocabulary development by showing that quantity of input is crucial very early on, while diversity of language — rare words and diversity of word types — has more predictive power in the later toddler years, with more complex decontextualized speech becoming indicative in the preschool years of a child’s vocabulary size at school age.

Studies like Rowe’s are giving us a complex picture of children’s language development early in life. It’s not just the number of words that children depend on to increase their vocabulary; children capitalize on different aspects of language at different ages to learn new words, and these relationships occur independent of their family's socioeconomic status. So, does this mean we need to just talk a lot to babies and then get more complicated as they get older? Chances are, you do that already. As your toddler or preschooler gets older, remember that they don’t just need to hear many words, but they need to hear a variety of words, some of which challenge them – for instance, by talking about events in the past or future, or imagining things that aren’t really there. You likely do this already if you live with a toddler or preschooler: Thinking about findings like these makes me more patient the hundredth time my preschooler wants to talk about what kind of animal she is today, or have an in-depth discussion of that one time we went to the bakery but they were out of treats. I know that the imaginary language and the discussion of past events may be seeding her language with more complex vocabulary that will serve her well later on.