# Neurobiology of language in autism: a systematic review of pediatric studies

## TL;DR

This review examines how brain imaging studies show varied language development in autistic children and infants at high risk for autism.

## Contribution

The paper provides a systematic review of functional neuroimaging studies on language in autism, highlighting methodological trends and gaps.

## Key findings

- EEG was the most commonly used neuroimaging modality in studies of language in autism.
- Infants at elevated risk showed reduced brain activity in temporal regions, which did not consistently persist into childhood.
- Atypical timing and magnitude of auditory and linguistic processing were observed in autistic children from infancy to late childhood.

## Abstract

Language abilities are heterogeneous in autism spectrum disorder. Functional neuroimaging offers critical insights into the developmental trajectory of language in autism, with the potential to inform tailored supports. Unlike previous non-systematic or modality-specific reviews, this systematic review aimed to identify developmental trajectories in autistic children and elevated likelihood infants across functional neuroimaging modalities. Across the included 31 studies, EEG was the most common modality (39%), followed by fMRI (26%), MEG (26%), and fNIRS (9%). Most studies (77%) used passive awake tasks; around half employed simple speech stimuli (55%) while the remainder used complex speech (45%). There was an underrepresentation of minimally and nonspeaking autistic participants. fMRI and fNIRS studies found that elevated likelihood infants displayed hypoactivation in bilateral temporal regions, which did not consistently continue into mid-to-late childhood. EEG and MEG studies indicate that the timing and magnitude of auditory-perceptual and higher-order linguistic processing remained atypical (e.g., delayed latencies, larger amplitudes) from infancy to late childhood. Across modalities, findings were sometimes inconsistent, particularly when accounting for variability in language abilities. We highlight a strong need for caution in generalizing findings and argue that future research should prioritize more diverse samples to better elucidate the neural trajectory of language in autism.

The online version contains supplementary material available at 10.1007/s11682-026-01140-y.