Autism and bilingualism: what the evidence says — and what it does not

Bilingual families are still sometimes told to choose one language after an autism diagnosis. The advice sounds cautious: perhaps two languages would be “too much.”

The evidence does not support that blanket restriction. A 2024 systematic review included 12 studies and 328 autistic children aged 3 to 12 and reported no additional language difficulty associated with bilingualism (Garrido, López & Carballo, 2024). A 2024 study of autistic toddlers likewise found no significant autism-by-bilingualism interaction for language, cognitive or social-emotional scores (Phillips et al., 2024).

Those findings are reassuring, not final. Samples are limited, definitions of bilingual exposure vary and individual language profiles are highly heterogeneous. “No evidence of harm” is not “every bilingual plan works equally well.”

The home language is more than input

A language belongs to jokes, comfort, grandparents, meals and the adults who speak most naturally in it. Asking a family to abandon that language can reduce the quality and quantity of interaction even if the advice was intended to simplify learning.

The useful question is not “which language should disappear?” It is “which people and situations use each language, and which relations does the learner need in both?”

Do not translate every trial mechanically

If a child learns spoken “dog” → one picture and then immediately receives spoken “perro” → the same picture with the same prompts, the program has taught two relations. It has not shown that either word works with a new dog, another voice or a real animal.

A stronger plan separates:

• teaching within each language;
• variation of speakers and examples;
• cross-language checks without help;
• generalization to objects and daily routines;
• maintenance after time.

That structure makes it possible to ask whether a relation emerged rather than counting translated repetition as proof.

Where Interlaza helps

Interlaza’s interface, family guidance and teaching content are available in Spanish and English, and every concept in the library carries a label in both. A pathway can be set to Spanish only, English only, or to alternate between the two from trial to trial — and the same concept can also be represented by realistic images, pictograms, spoken labels and written words, which is a separate axis from which language is in force.

That alternation is exposure, not a validated check: the record does not currently mark a bilingual trial as a no-feedback probe the way it can mark a held-out exemplar or a probe on other axes the app already tracks. Asking “does this carry from Spanish to English” as a real question means building that check by hand — holding one language’s label out of teaching and authoring a probe stage around it — the same way any other transfer axis needs its own stage rather than being assumed from an interface switch. None of this is limited to any particular tier; it is available wherever a pathway can be configured.

For families, the practical benefit is narrower and needs none of that: the adult does not have to operate an English-only tool while trying to preserve Spanish at home, or the reverse.

For Instructors, the record distinguishes an independent response from prompted practice, and — on any axis where a probe has actually been authored — a trained relation from a probe. The app organizes the plan; it does not decide which family language matters, and it does not claim to measure cross-language transfer on its own.

A respectful starting agreement

Ask the family:

1. Which language does each important person use most naturally?
2. In which language does the child currently understand or express particular routines?
3. Which words or relations need to work across both languages?
4. Which examples should remain new so transfer can be checked?
5. How will AAC, signs or other communication forms remain available in both contexts?

This conversation prevents “bilingual” from becoming a marketing checkbox. It becomes a description of the child’s actual communication world.