Word predictors work like a magic wand: you type the first three letters and the word appears. For a child with dysorthographia, that is tempting. But this technology raises a question that is rarely asked: at what cost? This article debunks the myths surrounding predictors and reveals why a “convenient” tool can become an obstacle to progress.
What word predictors promise (and what they do not do)
Predictors are everywhere: in phones, word processors, browsers. Their promise is clear: write faster and without errors. For dysorthographia, this is appealing. A child who mixes up verb agreements, reverses letters, or forgets silent letters suddenly gets correct spelling without effort.
Except that: getting correct spelling is not learning to spell. This is a major difference that is greatly underestimated. The predictor corrects the result, but not the mental representation of the spelling. The child does not visualise how the word is written; they just click, that’s all.
What actually happens: the child switches to “autopilot”. They type quickly, without thinking about spelling, because they know the predictor will rescue them. This is reassuring in the moment. But the brain has recorded nothing.
The trap of dependence: how predictors slow progress
Here is the paradox: the more a predictor is used, the more it is needed. Why? Because the child with dysorthographia stops becoming aware of their writing. They no longer try to memorise. They wait for the suggestion.
In practice, this is what happens:
- Phase 1 (weeks 1-2): relief. The child writes without panic, the errors disappear, the parents are pleased.
- Phase 2 (months 2-3): the predictor is no longer enough. Either it does not offer the right word, or the child types too fast and clicks without checking. Errors reappear, but worse: the child has forgotten how to search for the answer on their own.
- Phase 3 (beyond): frustration. Without the predictor, the child cannot write. With it, they depend 100% on the technology.
This is particularly true for children with dysorthographia, who need active repetition and self-correction strategies. The predictor removes the cognitive effort required for memorisation.
The real differences: predictor vs spell checker
Two things should not be confused:
- The predictor: it suggests words before you have finished typing. It is fast, but it eliminates reflection.
- The spell checker (such as Antidote or LanguageTool): it identifies errors after the fact and explains why. Far more instructive.
For a child with dysorthographia, the spell checker has educational value. It allows the child to write first, then correct while thinking. The predictor, on the other hand, removes this stage of reflection.
In our article on why digital tools are not enough, we explore in detail how technologies can create illusions of progress. Predictors are the perfect example.
When predictors suggest anything
There is another problem: predictors are not infallible. AI algorithms work on statistics. If you have typed “je sui”, the predictor will usually suggest “suis”. But on a poorly configured phone, or with odd language settings, it can also suggest bizarre words.
For a child with dysorthographia who has not yet internalised the correct forms, this is a problem. They may accept a wrong suggestion without knowing it. The brain retains what is seen and clicked, even if it is incorrect.
Concrete example: a child clicks “une” instead of “un” because the predictor offered it. Repeated ten times, this reflex becomes ingrained. A few weeks later, without the predictor, the error persists.
How to use a predictor without falling into dependence
This does not mean predictors should be banned. But they should be used strategically.
1. The predictor as a last resort, not a first reflex
Set a clear rule: the child first tries to write alone. The predictor is only activated if the child is genuinely stuck (after 30 seconds of thought, for example). In this way, the cognitive effort is maintained.
2. Check each suggestion, do not just click
Before accepting a predictor’s suggestion, the child should check mentally: “Do I recognise this word? Does the spelling seem right to me?” This takes 2 seconds, but it is the moment when the brain learns.
3. Turn off the predictor during consolidation exercises
If the child is working on specific points (agreements, homophones, endings), the predictor should not be active. This is the time to exercise the spelling muscle.
4. Combine predictor and spell checker
After writing with the help of a predictor, the child runs a spell checker (Antidote, BonPatron) to see what they missed nonetheless. This creates a feedback loop: the predictor helps; the spell checker teaches.
For younger children, apps such as dys’tap.io offer educational games adapted to dysorthographia, without a predictor but with solid teaching strategies. It is an excellent alternative for learning phases.
Predictive AI: the future, but not the solution
Predictors are becoming more and more sophisticated. Today’s AI systems can suggest entire sentences, not just words. ChatGPT and its equivalents can rewrite an entire text.
This raises a profound question: at what point does the child stop writing and simply start validating?
For a child with dysorthographia, this boundary is already blurred with simple predictors. With generative AI, it becomes almost invisible. The risk: the child moves from “someone who struggles to write” to “someone who entirely delegates to the machine”.
This is also what we explore in our article on how to enhance your child’s hidden strengths: accepting tools, yes, but not sacrificing autonomy and self-confidence.
Predictors: a professional tool, not an educational one
Here is an important nuance: predictors are excellent for productivity, not for learning.
- An adult with dysorthographia who is working? Yes, the predictor saves them time and stress. They do not need to progress; they need to produce.
- A child with dysorthographia who is learning? No. The child needs to build their mental representations of spelling. The predictor bypasses this process.
This is a distinction that schools and parents often overlook. We confuse “managing to write correctly” (thanks to the predictor) with “learning to write” (through conscious practice).
What to remember: predictors YES, but with limits
Predictors are not enemies. But they are not saviours either. Here is the summary:
- During learning: limit access to maintain cognitive effort.
- During consolidation: turn off completely.
- For “final” productions: use without guilt, but checking each suggestion.
- In parallel: always combine with conscious work (spell checker, targeted exercises).
How to help a child with dysorthographia at home without falling into common educational traps? Precisely by not substituting a technological tool for learning. Predictors are the classic example.
Frequently asked questions
Can a predictor improve my child’s spelling in the long term?
No, only conscious practice improves spelling. The predictor relieves the cognitive load, but does not embed knowledge. It must be combined with exercises and reflective correction.
Should predictors be banned at school?
Not banned, but supervised. Allowed for final productions, prohibited during learning phases. Many schools allow them at secondary level with adjustments, which is good practice.
Will my child with dysorthographia become dependent on technology if they use a predictor?
There is a risk, but no certainty. It all depends on how it is used. If they use the predictor as a permanent crutch without working in parallel, yes, they risk dependence. If they combine it with targeted exercises, it is one tool among others.
To find out more
- Ministère de l’Éducation nationale: resources on digital adjustments for children with learning difficulties and DYS disorders.
- Fédération Française des Associations de Dyslexie et Dysorthographie (3DFrance): guide to good practice with digital tools for dysorthographia.
- INSERM: studies on brain plasticity and the importance of active practice in learning spelling, even in children with DYS disorders.
