After about a year, give or take, of staring and babbling, babies eventually begin to say their first words. Mama. Ball. Dog. Millions of parents all over the world know this.

Now, researchers at Indiana University and the Georgia Institute of Technology have discovered new clues about how that actually happens—how babies learn those initial words. It turns out, the researchers report in a new paper, that a baby’s first words are likely tied to their visual experiences and how they see the world around them.

That might sound obvious, but Linda Smith, a professor of psychological and brain sciences at Indiana and senior author of the study, insists the work may provide the basis for new theories of how infants acquire language, and even how children with language deficits and autism are treated.

Researchers have known for some time that by about age 2, kids can use things like social cues and context to figure out which words and objects go together. By that age, toddlers are very good at grouping things into categories. In other words, if they know at home that a spoon is called a spoon, they will know that a spoon at a restaurant is also called a spoon, even if it doesn’t look or feel exactly the same.

But those skills don’t magically appear; they develop over time. And it hasn’t been clear exactly how a 1-year-old who doesn’t have those capabilities puts objects and names together. All that was clear was that the baby could—somehow. And while some insight has been gleaned by studying babies, sticking a kid in a sterile lab doesn’t exactly replicate the chaos and clutter of daily life.

So Smith and her colleagues essentially got on babies’ level to figure it out. They put head cameras on 8- to 10-month-olds to look at egocentric vision—basically, what each baby saw during the day. Many people assume, Smith told me during a phone conversation, that what babies and adults see is pretty much the same. After all, they’re living in the same houses, riding in the same cars, and going on the same grocery runs. But that turns out not to be the case, Smith said. Babies aren’t as good at controlling their bodies, and, when they can, they aren’t interested in looking at the same stuff as adults. As a baby develops, his visual world changes, Smith added, meaning a 3-month-old has an entirely different set of visual experiences than does a 1-year-old, who will have different experiences than a 2-year-old.

The researchers wanted to get a sense of the visual worlds of babies who are just shy of the age at which children utter their first words. They decided to focus on mealtimes, because babies eat multiple times each day, often in different contexts—say, in a high chair at breakfast and then in a stroller a couple of hours later on the way to the park. Ultimately, in a theory that Smith and her colleagues have dubbed the “pervasiveness hypothesis,” they found that objects that babies see most consistently are often the first words they say. In other words, babies see stuff all day long—toys and pets, tables and chairs, clothes and food. But what they look at often varies. There are a limited number of things—a bottle, for example—that babies see each time they eat. And “bottle” is more likely than the more seasonal “strawberry” to be among the first words a baby learns.

The researchers suggest that visual experience is the key element in learning first words. To make a word “stick,” it may matter more that a baby sees a bottle often than whether his parents are pointing at it and saying “bottle.” And the more often a baby sees that bottle, the more able he is to identify it in more visually cluttered situations. (If he sees it every day on an otherwise empty high chair, he’ll be more likely to recognize a bottle on a shelf next to diapers and teething rings at Target.) That could have implications for how kids with delayed speech and other disorders are treated: potentially through visuals rather than something else. If a kid has trouble learning words, he might have a visual-processing issue, or he might live in a house where he doesn’t see objects consistently.

Smith said the researchers cannot say for sure, but this research may mean that visual differences across how children experience life influence how quickly they begin to say words. A child living in an unstable situation, whose visual world is constantly shifting, may have a tougher time pairing objects with the words that identify them. There have been lots of studies around the “word gap” that exists between poor and wealthy children, which suggest that talking to babies early and often improves their chances of success in school and beyond—but there’s been less focus and clarity around exactly how babies learn their first words, and, Smith said, the visual elements have often been overlooked.

The study was supported by the National Science Foundation and is part of a larger effort to collect and analyze millions of images of what babies see in their first two years of life. The authors note that since this is the first study to look at how often babies see different objects (from their viewpoint, using cameras), the research raises a number of questions, including about individual differences. In a couple of months, the researchers will publish a paper looking at the frequency with which parents name certain objects and how that influences (or not) the acquisition of words. They also plan to look at how visual worlds are changing between 15 and 18 months, when children are learning words rapidly.

While some assumptions about how babies learn may be wrong, Smith said, she wouldn’t necessarily recommend altering daily life or trying to “de-clutter” what babies see. The take away, she said, is really that “your baby is always learning.”