Why your bilingual toddler invents different voices when she sings

If you have a bilingual or multilingual child somewhere between eighteen months and three years old, you may have noticed something charming and slightly odd: she sings the same song in different voices.

A deep grandfather voice with the cheeks puffed out. A thin, papery grandmother voice. The voice of a character from a book. The voice of an older cousin. Sometimes she announces it "now I sing like grandpa" and sometimes she just does it. And then, more remarkably, she will take a song she has only ever heard in one language and sing it in a voice that belongs to a different language entirely.

It looks like play. It is also one of the more sophisticated pieces of language work a small child does. Here is what is actually happening, and why bilingual children are particularly good at it.

What is happening: register variation

Around the age of two, children begin to develop what linguists call register variation — the conscious ability to change their voice to mark something about the speaker, the role, or the situation. The classic everyday example is a toddler who addresses a younger sibling or a doll in a higher-pitched, gentler voice than the one she uses with her father, then drops back into her normal register a few seconds later. She is not just imitating sounds, she is holding a different speaker in her mind well enough to perform their voice.

This is a developmental milestone. It maps closely onto the early stages of theory of mind, which is the slowly emerging recognition that other people have inner lives that are not the same as your own.

Strictly speaking, the full false-belief shift of theory-of-mind research happens later, around three to four years old. But the precursors — the modelling of others as separate beings whose voices and intentions you can hold in mind — appear from around two, and register-switching is one of the most visible signs that the machinery is online.

When a small child sings Wind the Bobbin Up in a deep voice, on purpose, and announces it, she is not performing mimicry: she is practising perspective-taking, in singing form.

Why bilingual children do this earlier

Bilingual and multilingual children tend to develop register variation a little earlier than monolingual peers, for a structural reason: they are already, by necessity, practising language-switching from the moment they have language at all.

A bilingual toddler has learned that:

• The same person can be addressed in two different ways depending on the language she is using

• The same chair, dog or biscuit has more than one name

• Which name to use depends on who is in the room

This early-internalised flexibility, the I can switch, does not only switch language, it also generalises to voices, roles, characters. The mental move of changing register to suit the listener is the same move whether you are changing from Italian to English or from your normal voice to grandfather's voice. The bilingual brain has been doing reps from the start.

Ellen Bialystok's long line of research into bilingual cognitive flexibility documents this effect across attention, switching, and metalinguistic awareness. The size of the so-called "bilingual advantage" is debated in the wider literature, but the narrower claim that bilingual children practise switching earlier and more often, is hard to dispute.

Why some languages "carry voice" more than others

You may also notice that your child's voice play is concentrated in one of her languages rather than spread evenly across all of them. This is usually not random. It is information about which language in her household carries the most prosodic distinctiveness, the most variation in pitch, timbre and stress.

Italian, Spanish, Portuguese and other Romance languages are what linguists call syllable-timed languages. Each syllable holds roughly equal duration; the music of a sentence comes from pitch and timbre rather than from the compressed rhythm of stress-timed languages such as English and German.

You can sing Italian in a way you cannot quite sing English. You can also tell whose voice an Italian sentence is in, even if the words don't change. Anne Fernald's foundational cross-linguistic work on infant-directed speech showed that Italian "motherese" carries a noticeably wider pitch range than English motherese. What a child's language exaggerates, the child notices. In other words, if voice and timbre carry more identifying information in one of her languages, that language becomes the natural ground for voice play.

The role of nursery rhymes

It is no accident that the material a small child chooses for voice play is so often nursery rhymes.

Nursery rhymes are built for prosody: they are constructed from rhythm, rhyme and repetition, which is exactly the material a young child's brain is most ready to work on. They are not great literature that's not how they have survived across centuries! But they tune the ear and give a child a framework on which to hang the strange and irregular shapes of her language.

Layering voice play on top of nursery rhymes is double work. You are practising the rhythm and rhyme of the song and you are practising the timbre and pitch of a particular speaker. For a multilingual child who is already constitutionally good at switching, this is the kind of layered exercise she will quite naturally reach for. Sandra Trehub's research on infant musicality at the University of Toronto, and Aniruddh Patel's work on the music-language overlap in the developing brain, both point in the same direction: music and language are processed in heavily overlapping networks in young children, and the songs that survive in childhood are the ones that train both at once.

What to do and what to avoid a parent

If your child is doing this, the most useful thing you can do is let her, and join in.

A few practical offerings:

• Encourage voice play of any kind. It is free, it is developmentally productive, and it is one of the lovelier things you will see this year. Ask her for the deep voice. Ask her for the thin voice. Sing along.

• Keep singing nursery rhymes long after she "knows" them. The repetition is the point, the voice play is the point.

• Don't correct the language switch. If she sings an English nursery rhyme in an Italian-coded voice, or vice versa, leave it alone. She has discovered that voice and language are decoupled, which is a piece of metalinguistic awareness that monolingual children take longer to arrive at.

• If a far-away grandparent, aunt or uncle has a memorable voice, ask them to record something. A short song, a lullaby, a snippet of conversation. Play it occasionally. You are giving your child more raw material for the voice-play she is doing anyway.

• Don't expect symmetry. Voice play tends to settle on the language that carries voice most distinctively. That doesn't mean the other languages in her life are weaker, only that her ear has picked the most legible voice-pair to work with.

A small closing

What looks like a toddler being funny in the bath is actually a toddler doing some of the most layered language work small children do — prosody, perspective-taking, switching, all at once. Notice it. Trust it. Sing along.

For the longer reflective essay on what this voice play taught a multilingual mother about the way absent grandparents arrive in a child's life, see the Sowing Tales piece Sing it like Nonno.