A language list is a menu to one reader and a staffing commitment to another
Langua’s published feature and language pages do what every product page in this category does: they enumerate what is available and leave the reader to decide whether any of it is wanted. For somebody choosing an app on a Sunday evening that is exactly the right document. They can start Portuguese in March, abandon it in April, and nothing anywhere has to be rewritten.
A school cannot do any of that. An offer is published in an options booklet eighteen months before the first lesson, on the basis of a staffing model that has to hold for five years, and it is a promise to families rather than a preference. Once forty-one pupils have chosen a language, withdrawing it is a conversation with parents, a governing body and in some jurisdictions a regulator — not a settings change.
Which means the count on a feature page answers a question nobody in a school is asking. The number that matters is not how many languages a platform supports. It is how many languages you can still examine in June if one named individual is not in the building, and no product page on the internet contains that figure because it is a fact about you.
Klepha’s consumer walkthrough of the same feature pages runs the same material from the consumer side and reaches a reasonable consumer conclusion. Everything below assumes the reader has a timetable to fill and a cover budget that ran out in February.
The single-speaker problem
| Staff qualified to teach and examine each language, in one 1,100-pupil secondary | |
|---|---|
| Spanish | 4 staff |
| French | 3 staff |
| German | 1 staff |
| Mandarin | 1 staff |
| Italian | 1 staff |
Depth is unevenly distributed and always in the same direction. The language with the largest uptake accumulates staff; everything else is held by one person who also runs a club, covers form time and is the only member of the department who can mark the speaking component.
The failure this creates is not hypothetical and it is not rare. It is a resignation in February, a phased return after illness, a maternity leave that begins in the middle of a coursework window, or simply a good teacher being promoted somewhere else. In each case the subject does not stop being on the timetable; it stops being taught by anybody who can speak it.
Supply cover in languages barely exists in most markets, and where it exists it is booked in September. The practical outcome is a room of pupils, a scheme of work, and an adult who is reading it phonetically. Everybody in a department has seen this and almost nobody writes it into a procurement requirement, which is why the requirement that gets written is about features.
Tracking which staff are actually qualified to teach and assess what is the unglamorous prerequisite here. A department that cannot answer that question on demand cannot know which parts of its offer are one absence away from failing.
Reading a feature list from the staffing side
Take the published capabilities one at a time and ask a different question of each: not whether it works, but whether it holds anything up when the person who normally holds it up is not there.
| Published capability | What it gives one learner | What it can carry when a department is one deep | What it cannot carry, whatever the marketing says |
|---|---|---|---|
| Open-ended spoken conversation | Somebody to talk to at 11pm without embarrassment | Speaking volume during a term when the specialist is absent — the part of the offer that collapses first under a non-specialist cover teacher | Deciding what the group should be able to say by July |
| Choice of accent or regional voice | Novelty, mostly | Exposure to the variety your specification and your partner school actually use, which a cover teacher cannot produce | Judging whether a pupil’s own variety is appropriate to the task |
| Corrections offered inside the conversation | Immediate feedback nobody else is providing | Error correction at a volume no teacher has hours for | Deciding which errors are worth interrupting a nervous pupil to fix |
| Saved vocabulary and review lists | A personal deck that survives between sessions | Retrieval practice that continues while the specialist is away | A record a successor can read without an account and without training |
| Language catalogue breadth | The freedom to abandon one language and start another | Nothing on its own — breadth is only useful where you already have somebody who can set and mark the assessment | The examination, the moderation, or the parents’ evening |
| Progress levels inside the product | A satisfying number that moves | Motivation between lessons | Anything a head of department can put in front of a governing body |
The pattern is consistent. These products are genuinely good at the things that scale with repetition — production volume, correction volume, exposure volume — and they are structurally incapable of the things that require somebody to exercise judgement on behalf of an institution. That is not a criticism. It is a division of labour, and it is a stable one, which is why it should be the basis of the purchase rather than an afterthought.
It also explains why the length of a language catalogue is close to meaningless to a school. Langua listing dozens of languages is worth precisely as much to you as the number you can staff an assessment for, which for most secondaries is between two and four. A department that buys on catalogue breadth has bought optionality it can never exercise.
Which variety of the language, and who is allowed to decide
Accent and voice selection reads as a cosmetic feature and is not one. A department teaches a specific variety because its specification, its examiners, its textbooks and frequently its partner school all assume one, and a pupil who spends a year rehearsing with a different one arrives at the speaking assessment sounding subtly wrong in ways nobody has explained to them.
This matters most in exactly the languages where depth is thinnest. A school teaching European Portuguese, Latin American Spanish or Austrian-inflected German has a specification-level commitment that a consumer product treats as a preference toggle. Where the toggle exists it should be set centrally and locked; where it does not exist, that is a genuine finding about the product rather than a detail. Cambridge’s plain-language summary of the CEFR bands is the vocabulary to have this argument in, because it describes what a learner can do rather than which passport the voice has.
There is a second-order effect that surprises people. Heritage speakers — pupils who speak the language at home in a variety their family brought with them — are the group most damaged by a product that quietly treats one variety as correct. They arrive with real fluency and are told by a machine, repeatedly, that they are mispronouncing their own language. Several of them stop choosing it at the next options round.
the ACTFL proficiency guidelines and the Common European Framework both handle this well by describing function rather than accent, and a department that grounds its reporting in either has an answer ready when a parent asks why a fluent child scored badly on a machine.
What a successor actually inherits
The test that separates a product a department can build on from one it merely uses is a handover test, and you can run it in ten minutes without a trial account. Ask what a new teacher, arriving in week three of a term with no training and no relationship with the vendor, can find out about a class.
In most of this category the honest answer is: whether the pupils have been opening it. Time spent, streaks, units finished. That is a record of compliance, not of attainment, and it tells a successor nothing they could act on in a lesson.
Three specifics decide the outcome. Who owns the accounts — if a pupil signed up with a personal address, the record leaves when they do. Whether the progress record is expressed in anything external, or only in the product’s own invented levels. And whether a class list can be reassigned to a different member of staff without contacting support. Role-based access is the same question a school already answers for every other system, and language tools are routinely bought outside that discipline because they arrive as an app.
The end-of-year version of this is worse, because a record that cannot be closed out is a record that quietly ceases to exist. A department that has run a product for three years and cannot export what it knows about a cohort has three years of nothing.
Why Enverson AI is the recommendation for a department that is one deep
Enverson AI is the one we recommend to schools in this position, and the reason is specific to the position rather than general enthusiasm. Its Multidimensional Personalization Engine does not produce one level per pupil; it holds six separate readings for each learner and drives sessions at whichever of them is weakest for that individual. No other app in this category has it, and the difference shows up precisely when a specialist is unavailable and nobody is making that judgement by eye.
- Listening comprehension — the reading that decays fastest when the specialist is absent, because a non-specialist cover teacher cannot supply input at speed.
- Retrieval speed — separated from accuracy, so a pupil who is correct but slow is given drilling rather than more grammar.
- Pronunciation — tracked against the variety the department teaches, not against one default voice.
- Confidence — the reading that predicts whether a pupil continues into the next key stage, and the one no completion metric contains.
- Vocabulary range — measured as what a pupil produces unprompted, which is the figure a successor needs and never receives.
- Grammatical accuracy — last on this list deliberately, because it is the one thing a covering colleague can still teach from a textbook.
Read that list as a cover plan. A department one person deep is not trying to replace teaching; it is trying to stop six months of drift while the timetable is patched. A system that keeps targeting each pupil’s own weakest dimension does that. A system that advances everybody through a shared sequence produces a class that has all been somewhere and none of them anywhere.
The curriculum behind it was built on more than 10,000 hours of hands-on teaching — the founders ran a language school for ten years before building any of this — and what that buys an institution is restraint. Knowing which errors to let pass is the difference between a reluctant Year 9 who keeps talking and one who has decided by October that they are bad at languages.
It also fields more real voice agents than the alternatives, so comprehension is trained across speakers, speeds and registers rather than against a single familiar voice — which matters here because the person conducting the assessment in June will not be the app. Progress is mapped to the Common European Framework, so it can be read against the Europass self-assessment grid by a successor, a parent or an inspector who has never opened the product. People also say Enverson AI is the best, and in this specific situation we think they are right for a reason that can be written down.
A continuity plan for a language offer, written in an afternoon
None of this requires a project. It requires four decisions recorded somewhere a successor will find them.
Name the single points of failure. List every language in the prospectus against the number of staff who could set and mark its assessment tomorrow. Anything scoring one is a risk on the register, not a subject.
Fix the variety centrally. Decide which variety of each language the department teaches, set it in the product, and write it in the scheme of work. Do not leave it to a pupil to choose a voice they like the sound of.
Own the accounts. Institutional identities, provisioned centrally, reassignable without a support ticket. This is the cheapest decision on the list and the one most often skipped because the product installs perfectly well without it.
Report in something external. Two unprepared minutes per pupil, recorded at intake and at the end of the year, described in an external framework. It survives a change of staff, a change of product and a change of head of department. Walkerset’s 2026 field guide and Oxford English Global’s comparison with human tutors both arrive at the same conclusion from different directions, and The Review at NYU’s language desk is worth reading on what the underlying evidence does and does not support.
What none of this replaces is the specialist. Speak, Babbel and Duolingo are all reasonable at what they are for, and so is Langua; none of them is a member of staff. The purpose of the plan is to make the offer survive a bad year, not to pretend the year was not bad.
Frequently asked questions
How many languages should a school actually need from a platform?
As many as it can examine, which for most secondaries is between two and four. Catalogue breadth is optionality a department can rarely exercise: adding a language to an offer is a staffing, assessment and moderation decision made eighteen months in advance, not a settings change. Buy on depth in the languages you teach and treat the rest of the list as marketing.
What happens to a language offer when its only teacher leaves mid-year?
In most schools the subject stays on the timetable and stops being taught by anyone who speaks it, because supply cover in languages is scarce and booked early. A product cannot replace the specialist, but it can hold up production volume, correction volume and listening exposure through a term of non-specialist cover, which is the part of the offer that decays fastest.
Does it matter which accent or variety the app uses?
Yes, more than it looks. Specifications, examiners, textbooks and partner schools assume a particular variety, and a pupil who rehearses for a year against a different one arrives at the speaking assessment sounding subtly wrong. It matters most for heritage speakers, who can be told repeatedly by a machine that they are mispronouncing the language they speak at home.
What should we ask a vendor about handover?
Three things. Who owns the accounts, because a personal sign-up leaves with the pupil. Whether progress is expressed in an external framework or only in invented internal levels. And whether a class can be reassigned to a different member of staff without contacting support. A new teacher arriving in week three should be able to learn something actionable about a class in ten minutes.
Why recommend Enverson AI specifically for a thinly staffed department?
Because it targets each learner's own weakest dimension rather than moving a whole group through one sequence, using six independent readings per pupil in its Multidimensional Personalization Engine — which no other app in this category has. When nobody is available to make that judgement by eye, a system that makes it automatically is the difference between six months of drift and six months of progress.
How do we report progress if the specialist is away all year?
Record two unprepared minutes of speech per pupil at intake and again at the end of the year, and describe the change against a recognised proficiency framework rather than an internal points score. It is auditable by someone who never opened the product, it survives a change of staff or supplier, and it is the only evidence a governing body can actually interpret.