New Fruits, aka "Rethinking the phonetics of baby-talk: Differences across Canada and Vanuatu in the articulation of mothers' speech to infants."

Babies and children are famously good at picking up the languages that they're exposed to; however, you've probably noticed that many people talk to babies very differently than they talk to adults.

A lot of people wonder whether this "baby-talk" (or "infant-directed speech") plays any role in helping children learn language early in life, and many research studies have described baby-talk in dozens of languages. Some have said that exaggerated, high pitch common in baby-talk is simply an emotional feature, that parents are just happy talking to their kids. Others have claimed that the high pitch and other features of baby-talk help to get babies' attention, which then in turn might help with language-learning.

Along with high pitch, short sentences, and slower speaking speeds, another feature of baby-talk that has been described in many languages has to do with how speech sounds are pronounced. Vowels in baby-talk are often pronounced with more subtle variability from word to word. That is, if you say "I saw three bees in the bee tree" to an adult, the "ee" sounds in that sentence were probably pronounced about the same as each other. However, if you say that sentence to a child, the "ee"s would probably each be a little more variable in pronunciation compared to each other, even the ones from the same word ("bee" and "bee"). You also might exaggerate or hyperarticulate the "ee" sound, making it sound very different from sounds like "ih" and "eh." Some people have said these features of baby-talk vowels, exaggeration and variability, make it easier for kids to learn to distinguish between different vowels. Some languages do the exaggeration (English usually among them), and so far all languages researchers have described have increased variability in baby-talk.

This project looked at how parents (particularly mothers) pronounced vowels in made-up words like "tisisi," "kususu," and "pasasa." The mother who participated were from Canada and Vanuatu, and spoke several different languages: all the Canadians spoke English, and the ni-Vanuatu mother spoke Lenakel, Southwest Tanna, and Weichen (languages which are part of the same dialect chain). The mothers were shown the four toys in the picture and told that they were newly invented fruits called tisisi, kususu, and pasasa. (The banana used its normal name in Canada and Vanuatu.) Moms were audio-recorded talking to their babies (6-20 months old) about the fruits, and after they'd said each of the made-up names at least 8 times, they were then recorded talking to an adult peer about them, again saying each word at least 8 times.

What we found

The researchers then picked out every time they said one of the made-up words in both recording sessions, and analysed each vowel to be able to describe their acoustic details. What we found was, when it came to these made-up words...

  • Both sets of mothers spoke in a higher pitch to their babies than to their friends/peers
  • Both sets of mothers spoke at about the same speed to their babies and their friends
  • Neither set of mothers pronounced their vowels in a more exaggerated way when talking to babies
  • Canadian mothers pronounced their vowels more variably when they spoke to babies than when they spoke to their friends, and each individual vowel in a word was less stable when speaking to babies (both findings were as predicted)
  • Ni-Vanuatu mothers surprised us: They made no difference between their babies and their friends when it came to vowel variability. This is the first time a version of baby-talk has been described that doesn't include increased vowel variability!
  • Ni-Vanuatu mothers also spoke with increased vowel stability (how much an individual vowel in a word changed from beginning to end) to their babies, compared to their friends, and in a less breathy voice; these facts were also surprising


This study left us with a lot of questions! More research needs to be done with languages from the same family as Lenakel, to tell whether this type of baby-talk is widespread. We're also interested in doing studies looking to see whether urban/rural living conditions make a difference in how people talk to their babies. No matter what, if the research community is going to understand more about how babies learn language, we absolutely need to learn more about what it is they're actually hearing all over the world, and not generalize too much while only having studied a tiny subset of languages.