Four on the FloorTechno for toddlers

The evidence

Every claim on this site, and what is wrong with it.

This is the page we would have wanted before we spent any money on our own child. It lists the studies we lean on, what each one actually did, and the reason each one is weaker than the headline it usually gets.

How to read this page

Each claim carries one of three labels. randomised means children were assigned to conditions by chance, so the study can speak about cause. observational means the researchers measured what families already did, so the study can only speak about association. guidance means a professional body issued a recommendation, which is a considered opinion and not a finding.

Nothing on this site rests on a single study, and nothing on this site is strong enough to justify spending money you do not have.

1. Turns predict how her language network develops

observational This is the claim the whole method rests on, so it gets the longest entry.

What the studies did. Home audio recorders captured the real language environment of 36 socioeconomically diverse children aged four to six. Children who had experienced more conversational turns with adults showed greater activation in Broca's area during a story-listening scan, and that activation statistically explained the link between their language exposure and their verbal skill. The effect held independently of family income, of the child's IQ, and of the sheer number of adult words.1 A companion study of 40 children in the same age band found that more turn-taking went with stronger, more coherent white matter connectivity in the left arcuate and superior longitudinal fasciculi, and specifically near Broca's area. That hypothesis was preregistered, and a whole-brain check found the relationship in no other tract.2

What it does not show

Neither study randomised anything, so neither can tell you that adding turns causes the brain to develop differently; families who converse more may differ in other ways. Both looked at children of four to six, not at babies, and this site is mostly about babies. Both are single samples of a few dozen children in one country. The honest sentence is that turns are the best-supported thing we know of to aim at, not that turns are proven to build tissue.

2. Coaching a parent changes what the child says

randomised This is the strongest evidence in the method, and it is still modest.

What the study did. 71 families of typically developing six-month-olds were randomly assigned to coaching or no coaching. Coached parents met individually after recordings at six, ten and fourteen months, heard their own audio, and were given feedback on how much they spoke directly to the child, how much back-and-forth there was, and how much parentese they used. Coaching significantly increased both parentese and turn-taking. At eighteen months the coached children produced more words in their own recordings, and their parents reported larger vocabularies.3 An earlier trial by the same group, with coaching at six and ten months, found more babble growth and more words at fourteen months.4

What it does not show

The effect on parent-reported vocabulary was small and only just crossed the usual threshold: t(68) = 2.03, P = 0.046, d = 0.53. The observed-speech measure was stronger, d = 0.78. The two groups were badly unbalanced, 48 coached families against 23 controls. Parents who knew they had been coached filled in the vocabulary questionnaire themselves, which is exactly the situation in which parent report flatters. And an eighteen-month vocabulary is not a school outcome; nobody has followed these children far enough to tell you it lasts.

3. Musical play in infancy shows up in how speech is processed

randomised The reason a record label thought it had any business here at all.

What the study did. Nine-month-olds were randomly assigned to twelve sessions of music activity or twelve sessions of play that matched the music condition for social contact, repetition and multisensory input but contained no music. The music sessions deliberately used triple meter, which is hard for infants. Afterwards, brain responses were measured with magnetoencephalography. The music group showed larger mismatch responses to timing violations, and the same result appeared for timing violations in speech, in a foreign syllable structure.5

What it does not show

The outcome is a neural response to a violated pattern, not a word, a sentence or a milestone. No one has shown that this particular gain turns into better talking, or that it survives a year. It was a small laboratory trial in one lab. What it does establish, and this is worth having, is that the direction of the arrow is real: music was assigned, not chosen, so the difference is not families who like music being different families.

4. A screen teaches only when it answers back

randomised This is why the interactive alphabet stops and waits, and why we would rather you used the book.

What the study did. 36 toddlers aged 24 to 30 months met new verbs in one of three conditions: a live person, a live video call, or a recorded video of the same session. They learned the words from the live person and from the video call. They did not learn from the recording, even though it contained identical language.6

What it does not show

It is one session, in a laboratory, teaching invented verbs. It is not a licence for unlimited video calls, and it says nothing about an hour a day of anything. Read it narrowly: contingency is the ingredient, and a recording has none.

5. Paper produces more talk than a tablet

observational Within-subject, which is stronger than it sounds, because each pair was its own control.

What the study did. 37 parent-toddler pairs were filmed reading three book formats in counterbalanced order: print, a basic electronic book, and an enhanced electronic book with sound and animation. Dialogic utterances by the parent, the kind that invite the child to say something, ran at 11.9 intervals with print against 6.2 with the enhanced book. Talk about the device itself ran the other way, 1.9 against 10.0.7

What it does not show

A single laboratory session with commercially available tablet books as they existed in 2019. It does not show that children learn less from screens in the long run, and it does not mean a device is harmful. The mechanism it exposes is duller than that: an enhanced book does the performing, so the adult stops.

6. The thirty-million-word gap, which we do not use

observational The most famous number in this field, and we have deliberately kept it off every other page.

What the challenge did. Five studies across American communities at different points on the socioeconomic range combined ethnographic fieldwork with longitudinal home observation of 42 children aged 18 to 48 months. The results do not support the claim that poor children hear thirty million fewer words. They also show wide variation within every social stratum, and they show that counting only speech addressed directly to the child by a primary caregiver, and ignoring other caregivers and overheard talk, undercounts low-income children's environments in particular.8

Why this matters here

A large part of the early-learning industry is built on that number. We think it is the wrong target twice over: it is contested empirically, and even at its best it points a parent at volume, which is the part of the effect that the turn-taking work found was not doing the work. So this site counts turns and never counts words, and now you know that this is a choice we made rather than a fact we inherited.

7. Screen limits are guidance, not findings

guidance We follow the guidance. We are not going to pretend it is something it is not.

The World Health Organization's recommendations for under-fives, including the sedentary screen-time limits, are labelled by WHO itself as "strong recommendation, very low quality evidence", and that same label is attached across every age band and to the physical activity and sleep recommendations too.9 The American Academy of Pediatrics sets out the familiar shape: video chat only below eighteen months, high-quality material with an adult present from eighteen to twenty-four months, and about an hour a day with an adult from two to five.10 The Royal College of Paediatrics and Child Health published its own guides for parents and clinicians in January 2019 and has since withdrawn them; its page now carries the line "these guides are no longer available", and the old address returns a 404.11 A body of that standing quietly retiring its screen-time advice is worth knowing about before you treat any number as settled.

Because we send people here from the alphabet page, here is the wording itself, transcribed from the two documents rather than summarised. WHO first, from the one-page recommendations:

Age bandWhat the WHO guideline actually says about screens
Under 1 “Screen time is not recommended. When sedentary, engaging in reading and storytelling with a caregiver is encouraged.”
1–2 “For 1-year-olds, sedentary screen time (such as watching TV or videos, playing computer games) is not recommended. For those aged 2 years, sedentary screen time should be no more than 1 hour; less is better. When sedentary, engaging in reading and storytelling with a caregiver is encouraged.”
3–4 “Sedentary screen time should be no more than 1 hour; less is better. When sedentary, engaging in reading and storytelling with a caregiver is encouraged.”

Then the American Academy of Pediatrics, from the recommendations to families in Media and Young Minds:

Age bandWhat the AAP statement actually says
Under 18–24m “Avoid digital media use (except video-chatting) in children younger than 18 to 24 months.”
18–24m “For children ages 18 to 24 months of age, if you want to introduce digital media, choose high-quality programming and use media together with your child. Avoid solo media use in this age group.”
2–5 “For children 2 to 5 years of age, limit screen use to 1 hour per day of high-quality programming, coview with your children, help children understand what they are seeing, and help them apply what they learn to the world around them.”

Two things are worth noticing in the exact words that get lost in the summaries. Both bodies put an adult in the sentence: WHO's alternative to the screen is reading and storytelling with a caregiver, and the AAP's allowance is high-quality material used together with your child. And the AAP's own limit is a limit on use, not a prohibition on the object. That is the whole reason the alphabet on this site waits for her and the book does not need a battery.

The neuroimaging that gets quoted alongside these limits is thinner than the coverage suggests. The best-known study scanned 47 children aged three to five and related a parent-completed screen-use questionnaire to white matter measures. It is cross-sectional, so it cannot tell you whether screens changed the brains, or whether families whose children were already developing differently used screens differently.12

What is not on this site because we could not stand it up

  • Right-brain training. A search of Europe PMC on 6 September 2026 for "right brain education" returned no records at all, and searches for the named commercial programmes return unrelated papers by authors who happen to share the name. We are not saying these programmes do nothing. We are saying that after looking, we could not find a trial to show you, and neither, as far as we can tell, can they.
  • Flashcards for babies. Same result: nothing we could cite.
  • Anything about music and IQ. There are claims. We are not making them.
  • Our own product. No one has ever run a trial on our book, and we will not imply otherwise. What is evidenced is the behaviour the book is designed to provoke, which is why the method is free and the book is optional.

What would change our mind

  • A well-powered replication of the turn-taking work in infants rather than four-to-six-year-olds. That is the load-bearing gap and we know it.
  • A follow-up on the parent-coaching trials at school age. Right now the trail goes cold at eighteen months.
  • Any decent trial showing that the musical effect reaches an actual language milestone rather than an evoked response.
If we have got something wrong

Write to us. We will correct the page, and we will leave a note saying what it used to say and who told us. Sources on this page were checked against Crossref and Europe PMC on 6 September 2026, and every link goes to the paper rather than to somebody's summary of it. Contact →

Sources on this page