The report comes home and there is one line in it you keep going back to. Does not initiate interaction with peers. Prefers parallel play. Requires prompting to greet familiar adults.
You read it three times. Then you picture your child last night, lying on the floor telling you in enormous detail about two species of shark, checking your face every few seconds to see whether you were still with them.
Both of those are your child. Only one made it into the report, and only one is what most social communication skills programs for autistic children are built to change.
Mind that gap before you sign up for anything.
How do I teach social communication skills to my autistic child?
Teach the skills that make communication work. Repairing a misunderstanding, asking someone to slow down, noticing your own overwhelm before it lands, saying no and having it stick. Then leave alone the ones that only make a child look less autistic. The test for any target: does this help my child get their needs met and connect with people they like, or does it mainly make them easier for other people to be around?
Those two overlap more often than either the therapy industry or its critics admit. They are still not the same thing, and when they conflict, the first one wins.
The double empathy problem, in plain terms
The standard story is that autistic people have a social skills deficit and the fix is to install the missing skills.
The double empathy idea, associated with the autistic researcher Damian Milton, reframes that. It proposes the breakdown between autistic and non-autistic people is mutual: a mismatch between two communication styles, not a one-sided failure. Non-autistic people misread autistic body language, tone and intent about as badly as the reverse. They just do not get told they have a disorder for it.
The picture emerging from research is broadly consistent with that. Autistic people talking to other autistic people tend to pass information along and build rapport much as non-autistic pairs do. Mixed pairs are where things break down. Researchers also report that non-autistic observers form negative impressions of autistic people very quickly, before much content has been exchanged.
The evidence here is younger and thinner than the older deficit literature, and people who dislike the framing will say so. Fair enough. Taken cautiously it still changes the job in front of you. Some of what looks like your child's difficulty is a two-way mismatch, and loading all the repair work onto the smaller, younger, less powerful party is unfair and inefficient. Teaching the people around your child counts as intervention, not as a cop-out. And a child with autistic friends may be socially successful in a way a standard assessment scores as failure.
None of that means nothing is worth teaching. It means being choosier about what.
What is genuinely worth teaching
Skills with a payoff your child will actually feel. They are also the ones most curricula spend the least time on.
Every one of these is a tool the child operates for their own ends. That is the pattern to look for.
Here is repair in practice, since it leads the list. Your child tells a classmate their drawing looks wrong. The classmate's face drops. You say nothing in the moment, because a correction delivered in front of a peer teaches shame, not language. Later, in the car, you name it plainly. "When you said the drawing looked wrong, she went quiet. I think she heard it as mean." Then you hand over the line: "You could say, that came out ruder than I meant." The following week, over something trivial at home, you ask them to try it once when nothing is riding on it. Repair gets taught after the event, never during it.
If you are working on the language underneath these skills, meaning words for feelings, another person's perspective, conversational turns, Verbalyft's story-based practice covers that ground without pressure, and ASHA's autism practice portal shows what a well-written social communication goal looks like.
What is really masking training
The other list. These show up constantly in programs, IEP goals and app curricula, and they are mostly about the comfort of observers.
Three more belong here without needing a bullet each. Requiring a child to greet every adult who speaks to them is a compliance goal in social clothing, because nobody greets everyone. Rehearsing questions about a peer's weekend is fine offered as an option and grim enforced as a requirement. And coaching a child out of a flat affect, or a big one, teaches them to perform feelings for an audience.
Beyond principle, masking has a cost. Autistic adults consistently describe long-term camouflaging as exhausting, and it is associated with burnout, anxiety and a weaker sense of self. A child praised for looking fine is often a child who falls apart at 4pm, and the falling apart gets treated as a separate problem.
Where the line is genuinely blurry
Ending there would be easy and dishonest, because plenty of real situations sit between the two lists.
Your child will at some point want a job, a friend who is not autistic, or entry to a group running on rules they find pointless. Refusing to teach any of it on principle is its own failure, and some autistic adults say exactly that. Nobody taught them the code, and they paid for it.
The way through is not to pick a side. It is to change the frame from rule to tool.
Then put weight on the half of the job almost no program does: teaching the environment. A teacher who allows fidgeting, a sibling told plainly that flat delivery is not rudeness, group-work rules written on the board instead of implied. Unglamorous, and it changes more than any curriculum.
Are social skills groups worth it?
Sometimes. The format has one real weakness: skills practised in a room with a facilitator often do not show up in the playground, and generalisation is where these programs fail. Questions worth asking before you pay:
| Ask this | A good answer sounds like | Walk away if |
|---|---|---|
| What are the goals? | Self-advocacy, repair, friendship the child wants | Eye contact, quiet hands, compliance |
| Who is in the group? | Children matched by interest and communication style | Ages 6 to 12 lumped together |
| Are autistic adults involved in the design? | Yes, and here is how | Blank look |
| What happens outside the room? | Coaching in the actual playground or classroom | Nothing, it is all in-session |
| What if my child does not want to come? | We change what we are doing | We work through refusal |
The strongest alternative is often not a group at all. It is one activity built around a genuine interest, whether that is chess club, a coding class or Lego, where interaction happens as a side effect of something your child came for.
What to say instead of "look at me when I'm talking"
Small substitutions, used often, do a surprising amount.
| Instead of | Try |
|---|---|
| Look at me when I'm talking | I'll wait. Tell me when you're ready |
| Use your words | Tell me any way you like |
| Quiet hands | Do you need something to hold? |
| Don't be rude | I don't think they knew you were joking. Want to try again? |
| Stop talking about sharks | Two more minutes on sharks, then I want to tell you something |
| You're being too loud | It's loud in here for me. Can we go quieter for a bit? |
The last two are doing something specific. They model repair and state-naming in front of your child, from you, at low stakes. That is how most of this is learned. By watching a trusted adult do it badly and fix it, not in a session.
FAQ
My child talks at people rather than with them. How do I teach conversation?
Start from a topic they already care about rather than a turn-taking drill, because a bored child will not practise. Model the swap out loud with the shark line above: two more minutes on their subject, then a short turn on yours, then back. Over months that becomes a rhythm they can run themselves, which is a different thing from memorising three polite questions.
My child has no friends and does not seem to mind. Should I intervene?
Take the not minding seriously before you act on it. Some autistic children genuinely prefer solitude or one close friend, and manufacturing a social life serves the parent's anxiety more than the child's needs. The signals worth acting on are distress, exclusion the child notices, or wanting friendship and failing to get it.
How do I get school to stop using compliance-style goals?
Ask for the goal to be rewritten around what the child gains rather than what the adult observes, and ask for it in writing. "Will remain seated with quiet hands" becomes "will use a break card to leave and return." Same behaviour, different owner, and much harder to argue against in a meeting.