New Tool Identifies Which Children Need Speech Therapy - First Updated English Data in Over 20 Years

Nearly every parent of a three-year-old has heard speech that sounds unusual - transpositions, substitutions, sounds that simply are not there yet. Most of the time, these errors are developmental. They resolve without intervention as the child's nervous system matures and they accumulate more practice with language. But a meaningful fraction of children have speech errors that will not resolve on their own, that signal an underlying disorder requiring therapy, and that worsen in impact the longer they go untreated.

The problem for clinicians and families is distinguishing one from the other. Without reliable reference data on what typical speech development looks like at each age - and without a screening tool calibrated to the full range of normal variation - the result is either over-referral (sending children with typical development to speech therapy, consuming limited clinical resources) or under-referral (missing the children with genuine disorders who need early intervention most).

A study published in the Archives of Disease in Childhood by researchers at Murdoch Children's Research Institute (MCRI) in Melbourne addresses both problems. It provides updated reference data on speech development and a practical screening tool built from those data. The last comparable English-language reference data was published more than two decades ago.

What 1,179 children's speech data shows

The study recruited 1,179 participants aged 2 to 12 years from schools, childcare centers, and kindergartens across Victoria and New South Wales. Trained speech and language therapists assessed each child using a standardized picture-naming task - a method that elicits a defined set of speech sounds across a range of positions in words, allowing systematic comparison across children and age groups.

The headline finding is straightforward: by age seven, 90 percent of children could correctly produce all sounds in the English phoneme inventory. From ages eight through twelve, only minor speech differences remained, and the overall error rate was low.

Before age seven, however, the picture was considerably more variable. Developmental speech errors - the kind that typically resolve - were common from ages two through six. These include familiar patterns like saying "wabbit" for rabbit or "tat" for cat, where the articulatory system is still maturing. The study confirmed these patterns are prevalent and should not be treated as signs of disorder.

Separate from these developmental errors, the researchers identified a category of disordered speech errors that occurred in fewer than 10 percent of children at any age. These included vowel errors, transpositions (such as "efelant" for elephant), and atypical sound substitutions (such as "glack" for black). The rarity of these patterns at any age distinguishes them from developmental norms and provides the diagnostic basis for identifying children who need attention.

Speech errors are taking longer to resolve

Comparing the new data to the historical reference data published more than 20 years ago revealed a troubling pattern: some sounds are now being acquired more slowly, and some common errors are taking longer to disappear, compared to the earlier cohort. The researchers emphasize that this does not indicate children's speech has become more disordered - the type and nature of errors look similar. But the timeline is different.

Possible contributors include changes in the ambient language environment - more screen time, potentially less adult-to-child direct speech, or changes in how language input is delivered during the critical early years. The study does not resolve this question, but Professor Angela Morgan, MCRI Group Leader for Speech and Language, noted that "new data is also needed to find out how new technologies, like phones and devices, are changing children's speech."

The screening tool and its intended use

The MCRI researchers developed a screening tool that uses the picture-naming task data to identify children who fall outside the normal range for their age group - specifically, those showing error patterns that occur in fewer than 10 percent of same-aged peers. The tool is designed for use by pediatricians, speech therapists, and other healthcare professionals who encounter children in early childhood settings.

Its value is primarily in triage. Rather than requiring all children with any speech errors to be assessed by a specialist, the tool helps prioritize those whose patterns suggest genuine disorder while providing reassurance that children showing typical developmental errors are likely to catch up without intervention.

Wait times for speech therapy are long in many health systems. Resources directed toward children who would resolve without treatment are resources unavailable to children who will not. The screening tool is explicitly designed to improve this allocation.

Researchers from the University of Melbourne and Redenlab also contributed to the study.