Teacher guide

You will find it in the app menu under 🧩 Inclusive teaching. It is not an on/off switch, and that is deliberate: what a dyslexic student needs is not what a student with a cognitive disability needs, nor what someone with an attention disorder needs. A single “simplified mode” would give everyone a mediocre compromise.

So you pick the aspects that student needs, and how strongly.

Intensity

The three levels are not a label: they genuinely change the instructions. At light sentences stay under twenty words, at medium under fifteen, at strong under eleven and subordinate clauses are avoided too. The thresholds follow the European easy-to-read guidance, which puts the ceiling for accessible text at about fifteen words. Start at medium and adjust after seeing a real answer.

It is not only about shorter sentences. The change that matters most is the order: the explanation is reorganised around what happens rather than around scientific classification. On photosynthesis, for instance:

The correct technical names remain, but they arrive after the thing has been explained, not instead of it. Other things change along with the order: passive becomes active, metaphors disappear because they are taken literally, technical terms are explained on the spot, and a concrete example comes before the rule. With comprehension checks on, each concept is followed by a short question, and a wrong answer leads to a different explanation, not the same one repeated.

Simplifying does not mean impoverishing, and Minerva is explicitly told not to: the content stays correct and complete, necessary steps are not skipped, what changes is how they are presented. It is also instructed to keep a tone that is never patronising: the student has their own age and is treated accordingly, with no diminutives and no remarks about how hard the task is.

It remains a tool, not a diagnosis or an education plan. The teaching decisions are yours: Minerva carries out the adaptation you specify.

1. Upload the syllabus or handouts in your area, under “Your materials”. From then on explanations and tests align with your course rather than with generic knowledge: same terminology, same topics, same angle.
2. Prepare the explanation for the class with inclusive mode off, at the depth you need.
3. Turn on inclusive mode and ask the same thing: you get the adapted version, with the same sequence of concepts. It is the work you would otherwise do by hand every afternoon.
4. Export both versions to PDF from the button under the answer, and take them to class.
5. Generate a test on the topic, timed or open: the assessment explains the mistakes instead of merely counting them.

It does not replace the educational relationship and it does not know the student: what blocks them, what motivates them, how they are today. You know that, and it remains the part that matters. Minerva takes time away from preparing materials to give it back where it counts.

Nor is it a source to quote unchecked: answers should be read before handing them to a student, as with any material found elsewhere. There is an independent check feature that reviews the answer and flags errors, and it is recommended when preparing material to hand out.

The adaptations are not invented: they follow established findings from educational and psychological research. It is worth knowing which, because they also explain the choices that might surprise you.

Design for functioning, not for diagnosis

This is the core of CAST's Universal Design for Learning, the most widely used framework in the field. Its premise is that “learner variability is predictable” and that “there is no average brain”: support should be organised around ways of functioning, not around diagnostic categories. Two students with the same diagnosis have different needs, and the same reading difficulty appears in dyslexia, in ADHD and in a student working in a second language.

That is why the configuration window asks which difficulty you are addressing, not which diagnosis the student has. Each profile still says who it typically helps, so you recognise your case without the app recording anything clinical.

There is also a legal reason, and not a minor one: a diagnosis is health data under Article 9 GDPR, concerning a minor and held in someone else's account. Minerva stores none: only the set of adaptations you chose is saved.

Words and pictures together

Research on multimedia learning, notably Richard Mayer's, shows that learning is better when verbal and visual material arrive together rather than apart. That is why the image is placed right after the sentence it belongs to rather than at the end, and why inclusive mode uses many more of them.

The same line of work warns that extraneous or decorative material harms learning instead of helping. So Minerva is instructed to request precise, recognisable subjects — a work, an organ, an instrument — and to insert nothing when no relevant image exists: better no picture than any picture.

One step at a time

Splitting into numbered steps, with a pause and a summary before the next, addresses the limits of working memory: present information in manageable chunks rather than as a continuous flow. It is the same reason why, in the attention profile, each step closes with a bold one-line summary: whoever loses the thread picks it up without rereading everything.

Literal does not mean poor

Banning metaphors, irony and implications mainly concerns those who tend to interpret literally, a frequent trait in autism spectrum conditions. In that profile Minerva announces up front how many steps there will be and in what order, then follows exactly that order: predictability lowers the load for anyone who struggles with the implicit.

Simplify the form, not the content

This is the distinction that holds everything else together, and the one Italian regulation insists upon, separating compensatory tools, which change the how, from dispensative measures, which remove. Minerva does the first job: it has explicit instructions not to omit necessary steps and to keep the content correct.

References: CAST — Universal Design for Learning; research on multimedia learning (R. Mayer); Italian Law 170/2010 and its ministerial guidelines on specific learning disorders. Minerva is neither a certified nor a clinically validated tool: it applies these findings to how answers are written, and does not replace professional assessment.

In the configuration window, above the aspects, you will find “Not sure where to start?”. It opens thirteen questions about what happens when the student studies, in five areas: reading (rereads the same line, is put off by a dense page), attention and memory (loses the thread, drifts off after a few minutes, has lost the beginning by the end), language (takes idioms and irony literally), numbers and procedures (mixes up digits, skips steps, does not notice an absurd result), understanding (reads but cannot say what they read, does not know many everyday words, needs an example first, quietly accumulates gaps). For each you answer never, sometimes or often.

Before anything is applied you see the outcome in plain words: which adaptations switch on, at what intensity, and which starting point is suggested. From there you can go back to the questions or accept. It is a starting point, not a verdict, and it stays editable.

A single “sometimes” turns nothing on: it takes a frequent observation, or two different ones converging on the same need. That is deliberate — everyone has the occasional difficulty, and adapting everything amounts to adapting nothing.

If you follow several students you can save the configuration under a reference of your own — “3A-07”, “Wednesday” — and recall it when needed. Use a code, not a name: Minerva keeps only the reference you type and the set of adaptations, with no clinical data attached to it.

A fair question: do the observation questions meet the criteria a psychologist would apply to an instrument? Partly yes and partly no, and it is right to know where the line falls.

What it does meet

The questions are about observable behaviour, not traits of the person: “rereads the same line”, not “is dyslexic”. That is the form of teacher observation questionnaires, and the logic of the UDL framework, which designs on functioning rather than on labels. The three-level frequency scale (never, sometimes, often) is the usual one. The questions ask you to think of the last few weeks and to compare with classmates of the same age: without a time frame and a reference, “often” means nothing. And you are reminded to consider simpler explanations first — what manuals call exclusion criteria: eyesight or hearing to be checked, a second language, long absences.

What it does not meet, and does not claim to

There is no normative sample: nobody measured how a student population answers, so there is no score “above the norm”. The activation thresholds are design choices, not cut-offs derived from data. There are no reliability or validity measures: nobody checked that two teachers answer the same way about the same student, nor that the outcome predicts anything. It is not calibrated by age, and it rests on a single observer. A real screening questionnaire does all of this, which is why screening is administered, not downloaded.