Investigating language impairment in autism

All children with autism display some degree of language impairment, falling short of milestones related to producing and understanding language at the expected time. About one-third of children with autism never develop speech and remain mute throughout life. In those children with autism who do have speech, verbal abilities range widely.

Understanding the nature of the language impairment in autism and how variations in the pattern of early speech development might correlate with later language function are goals of a project "Precursors to Language in Autism," led by Dr. Patricia Kuhl.

In an approach new to autism research, Kuhl, professor of speech and hearing sciences, is using five detailed measures she has refined during 20 years of research in typical language development. The measures test basic abilities that underpin language. She uses them to determine whether language deficits in autism are caused by a fundamental problem in language processing or are a result of social deficits, which may curtail language learning experience. The measures also provide the basis for assessing relationships between patterns of early language development and later language outcomes, ­crucial information for guiding intervention efforts.

A fundamental aspect of language development is the ability to differentiate speech sounds. To determine if children with autism can hear the difference between simple phonetic units, Kuhl is using event-related potential (ERP) techniques. She is looking at a particular measure of electrical activity in the brain known as mismatched negativity (MMN). This measure takes a distinctive shape on the graph of the electrical signal in the brain when it detects a change in stimulation. The first test in the study involves a voice repeating a string of syllables with an occasional syllable that deviates from the one being repeated. For example, the child might hear "ba, ba, ba," then "da," then back to "ba, ba, ba." The presence of MMN reflects whether the child's brain distinguishes between the two syllables.

It's been traditionally believed that children with autism do not show deficits in discrimination of basic speech sounds. Their language impairment is thought to arise instead from another aspect of fundamental processing: -they may be unable to form language categories. This ability typically develops over time as a result of experience. Her work has shown that typically developing young babies can discriminate the phonetic building blocks of all the world's languages, but by 6 months they focus on the sounds of their own language.

Typically developing children listen to the language spoken around them and unconsciously their brains become "wired" to distinguish what they are hearing, says Kuhl. But this process may not occur in autism. Deficits in social cognition, which limit an autistic child's interest in people, may interfere with the experience necessary to lay down the wiring. The second test taps the effect of linguistic experience by measuring reaction to prototypical sounds from different language categories.

The third test is examining echolalia, a persistent repetition of sounds exhibited by some children with autism, particularly those whose symptoms are most severe. It is not known why children with autism produce echolalia.

Imitation is a major part of typical development, but echolalia is different. By looking at echolalia in a formal measurement situation for the first time, the study to begin to understand what function it serves for children with autism or whether, as it seems, it is the tendency to imitate at random.

The final two tests in Kuhl's study rely on preference measures. One assesses whether a child with autism is able to correlate the sight of a person's face while speaking with the sound of a voice. The other documents what kinds of sounds a child with autism would rather listen to. Some evidence suggests that children with autism prefer sounds made by objects rather than language sounds.

Each of the five measures taps a different kind of language ability, ranging from basic perception of sounds to higher levels of cognitive processing. Combined, the measures make a language profile of each child that answers fundamental questions about language impairment in autism. And, in conjunction with other measures from the longitudinal study, these profiles are deepening understanding of autism in general.