Can a Mother's Recorded Voice, Switched On by Her Baby, Help Preterm Babies Learn to Listen and Talk?

What the MIND trial by Maitre and colleagues (Journal of Pediatrics, 2026) found about babies' brain responses to speech and their language at age two

In a study of 242 preterm babies in three U.S. hospitals, babies who were held by a therapist and could "switch on" their mother's recorded voice by sucking on a pacifier showed clearer brain responses to different speech sounds before going home. At age two, average language scores were almost the same in both groups; a benefit appeared only after taking into account how early each baby was born and how many years of school each mother had, so the result is encouraging but not yet settled.

Why language is a worry for families of babies born early

Many parents of babies born early have the same quiet worry: will my child learn to talk and understand as well as other children? It is a fair question. Babies born early are more likely to find language harder, often in ways that only show up later. At the same time, the weeks in a newborn intensive care unit (NICU) fall during a busy period of growth for the parts of the brain that handle sound. Parents naturally wonder whether there is anything that can be done during those weeks to give their baby's listening brain a better start.

This article looks at one carefully designed study that tried to answer that question. It was called the Multisensory support to Improve Neural processing and Development (MIND) trial, and it took place in three NICUs in the United States: Nationwide Children's Hospital in Columbus, Ohio, and Emory University Hospital Midtown and Grady Memorial Hospital in Atlanta, Georgia [1]. "Multisensory" simply means that the programme involved several senses at once: hearing, touch and smell.

How doctors and families have approached this problem until now

For many years, researchers have known that children born early tend to score lower on language tests, and that some of these differences grow rather than shrink between the ages of 3 and 12 [2]. A large Norwegian study that compared brothers and sisters found that children born before 29 weeks of pregnancy were much more likely to have delayed language at 18 months, although the extra risk became smaller by the time they started school [3]. Differences can already be seen in the first 18 months of life [4].

Part of the reason may be what babies hear in hospital. When researchers recorded the sounds around very small babies for 16 hours at a time, grown-up talk made up only a small part of what the babies heard, and babies had more back-and-forth "conversations" when a parent was there [5]. So hospitals have worked to help parents spend more time with their babies, although how much time parents can be there varies a great deal from family to family [6]. Many units also began playing recordings of the mother's voice. A review of these studies found that the methods differed widely, but that a mother's voice appears to be safe and may help with feeding and early development [7]. Other studies tried more active approaches. In Finland, parents who were coached to sing during skin-to-skin cuddling had babies whose brains responded more strongly to changes in speech sounds by their due date [8]. In Providence, Rhode Island, a programme that taught parents to talk more with their babies in the NICU was linked to fewer children with low understanding-of-language scores at age two [9].

The MIND researchers had also spent years measuring how babies' brains respond to speech sounds such as "ba", "da" and "ga". In earlier work they found that these brain responses, recorded before babies went home, helped predict language skills at age two [10]. In a small related trial at Nationwide Children's Hospital, babies who could "ask" for a recording by sucking on a special pacifier showed brain changes linked to the language they heard [11]. The open question was whether this kind of baby-led listening, given by trained staff when parents could not be there, would make a lasting difference.

What the study did

Babies could join if they were born before 36 weeks of pregnancy, had parents who spoke English and were medically stable once they reached 32 to 36 weeks, counted from the start of pregnancy. Babies who needed a breathing tube, had serious birth differences, had a known genetic hearing loss or were on sedating or seizure medicines could not join. Babies with some brain injury were welcome.

Each mother was recorded talking, reading and singing in a warm "baby-talk" style with help from a music therapist. The babies were put into one of two groups by chance, like the toss of a coin. In the comparison group, nurses played the mother's 20-minute recording once or twice a day, and the baby simply listened. In the MIND group, a trained therapist held the baby against their chest in a soft wrap, with the baby's head resting on a cloth that carried the mother's scent. Each time the baby sucked on a special pacifier, it played 10 seconds of the mother's voice, so the baby was in charge of when it heard her. Sessions lasted 20 minutes and stopped if the baby fell asleep, lost interest or showed signs of stress such as pauses in breathing. MIND babies had about 17 sessions on average, and comparison babies about 22 listening sessions.

Before and after the sessions, the team measured the babies' brain responses to speech sounds using a soft cap of sensors. This is called an EEG (electroencephalogram), and it is painless. At about two years of age, the children's language was tested with two standard sets of games and questions that check how well a child understands words and how well they express themselves.

What the researchers found

Of the 242 babies, 198 had usable brain recordings both before and after (100 MIND, 98 comparison). Babies in the MIND group showed clearer differences in their brain responses to sounds like "ba" and "ga" in several areas of the brain, though not in every area or for every pair of sounds tested [1]. Babies in both groups improved. The researchers think hearing the mother's recorded voice may have helped too, but because every baby heard the recordings, the study cannot tell this apart from babies' brains simply maturing with time.

At age two, about 169 children were tested. Here the picture is more mixed. If you simply compare the average scores of the two groups, they are almost identical: for example, 7.0 versus 6.7 on one understanding-language test, and 7.3 versus 7.4 on a speaking test (MIND group first, comparison group second). On a second set of tests the averages were also very close, around 89 to 91 points in both groups. The researchers then took into account two things that can affect language: how early each baby was born, and how many years of school the mother had completed. By chance, mothers in the comparison group had about one more year of schooling on average (14.7 versus 13.7 years). After this adjustment, the researchers reported a small-to-medium benefit for the MIND group on all four language scores, which they describe as about half a standard deviation (a statistical yardstick for how much scores usually vary between children). That benchmark for a "meaningful" change was borrowed from studies of quality-of-life questionnaires used in other health conditions, not children's language tests [12].

The researchers also used statistical models to argue that the brain changes in hospital led to the better language later; they describe the effect as "causally mediated". The overall effects in these models were statistically clear in only a handful (5 of 48) of the combinations they tested, and the detailed results that would show whether the brain changes really carried the benefit are in a supplementary file that could not be obtained.

What this means for your baby's care

This is careful, encouraging research, but it is an early step rather than a final answer. It suggests that babies, even very small ones, can take an active part in what they hear, and that this can change how their brains respond to speech before they go home. It does not yet show that the programme clearly improves how well children talk at age two. Because MIND combined several things at once (being held, the mother's scent, the therapist's calm and the baby-activated voice), nobody can yet say which part mattered most. The programme was delivered by specially trained therapists as part of a research study. It is not something to try to recreate at home, and it is not a replacement for you.

The most reassuring message is the simplest one. The researchers themselves see a parent's own presence, skin-to-skin cuddling and live talking and singing as the ideal. When you can be there, your voice matters. When you cannot, ask your NICU team what support they offer, such as recordings of your voice. No baby in the study failed a hearing test over two years, and sound levels during the sessions were kept gentle.

What researchers are working on next

The same team has since run another multisite U.S. trial testing only the baby-activated voice, without the holding or scent, and found a modest improvement in brain responses [13]. Future studies will need to work out which parts of the programme help most, include families who speak other languages, and follow children for longer than two years. If you want to read more about the sound environment of the NICU, see our related article, Noise, Silence, and Meaningful Sound.

In summary

The MIND trial shows that a short, gentle programme in which preterm babies switch on their mother's recorded voice while being held can sharpen how their brains respond to speech before discharge. Whether it improves language at age two is still uncertain, because the average scores looked the same until the researchers took into account how early each baby was born and how many years of school each mother had. It is a hopeful sign for future care, and a reminder that your own voice and presence remain at the heart of your baby's development.

References

  1. Maitre NL, Kjeldsen CP, Jeanvoine A, Lukemire J, Slaughter JL, Key AP. The MIND Randomized Controlled Trial: An Intervention to Improve Neural Speech Processing and 2-Year Language Outcomes of Infants Born Preterm. J Pediatr. 2026;296:115187. doi:10.1016/j.jpeds.2026.115187 ↩
  2. van Noort-van der Spek IL, Franken MC, Weisglas-Kuperus N. Language functions in preterm-born children: a systematic review and meta-analysis. Pediatrics. 2012;129(4):745-754. doi:10.1542/peds.2011-1728 ↩
  3. Zambrana IM, Vollrath ME, Jacobsson B, Sengpiel V, Ystrom E. Preterm birth and risk for language delays before school entry: A sibling-control study. Dev Psychopathol. 2021;33(1):47-52. doi:10.1017/S0954579419001536 ↩
  4. Loeffler MT, Daikeler J, Wagner L, Natalucci G, Daum MM. Early Expressive and Receptive Language Development in Preterm vs Full-Term Children: A Meta-Analysis. Pediatrics. 2025;156(2):e2024070477. doi:10.1542/peds.2024-070477 ↩
  5. Caskey M, Stephens B, Tucker R, Vohr B. Importance of parent talk on the development of preterm infant vocalizations. Pediatrics. 2011;128(5):910-916. doi:10.1542/peds.2011-0609 ↩
  6. Powers SA, Taylor K, Tumin D, Kohler JA Sr. Measuring Parental Presence in the Neonatal Intensive Care Unit. Am J Perinatol. 2022;39(2):134-143. doi:10.1055/s-0040-1715525 ↩
  7. Provenzi L, Broso S, Montirosso R. Do mothers sound good? A systematic review of the effects of maternal voice exposure on preterm infants' development. Neurosci Biobehav Rev. 2018;88:42-50. doi:10.1016/j.neubiorev.2018.03.009 ↩
  8. Kostilainen K, Partanen E, Mikkola K, et al. Repeated Parental Singing During Kangaroo Care Improved Neural Processing of Speech Sound Changes in Preterm Infants at Term Age. Front Neurosci. 2021;15:686027. doi:10.3389/fnins.2021.686027 ↩
  9. McGowan EC, Caskey M, Tucker R, Vohr BR. A Randomized Controlled Trial of a Neonatal Intensive Care Unit Language Intervention for Parents of Preterm Infants and 2-Year Language Outcomes. J Pediatr. 2024;264:113740. doi:10.1016/j.jpeds.2023.113740 ↩
  10. Maitre NL, Lambert WE, Aschner JL, Key AP. Cortical speech sound differentiation in the neonatal intensive care unit predicts cognitive and language development in the first 2 years of life. Dev Med Child Neurol. 2013;55(9):834-839. doi:10.1111/dmcn.12191 ↩
  11. Richard C, Jeanvoine A, Stark AR, Hague K, Kjeldsen C, Maitre NL. Randomized Trial to Increase Speech Sound Differentiation in Infants Born Preterm. J Pediatr. 2022;241:103-108.e3. doi:10.1016/j.jpeds.2021.10.035 ↩
  12. Norman GR, Sloan JA, Wyrwich KW. Interpretation of changes in health-related quality of life: the remarkable universality of half a standard deviation. Med Care. 2003;41(5):582-592. doi:10.1097/01.MLR.0000062554.74615.4C ↩
  13. Kjeldsen CP, Moran M, Jeanvoine A, et al. Recorded Contingent Caregiver Voice Independently Improves Neural Speech Processing in Hospitalized Preterm Infants: A Multisite Randomized Controlled Trial. J Clin Med. 2026;15(17):6880. doi:10.3390/jcm15176880 ↩