Linguistic · The Title III EL program-management surface · Early access · 2026
A fail-closed home-language-survey screen-before-services fact, a committee-facing identification recommendation from the language screener and the state’s threshold, an owed-services and reclassification workflow built on the canonical exit gate, and a consent chokepoint over EL-subject PII. It is deterministic decision support with no AI — the committee decides every classification and every exit. The engines are built; there is no live SIS sync and money is honest-off. Book a conversation and see the current state honestly.
How it works
From the home-language survey to the reclassification recommendation. Every step is described as it is built today, and the committee holds every determination.
Step 1 · Intake — the home-language survey and the screen-before-services fact
A coordinator brings the home-language survey answers into the worksheet. Any answer that is not an explicit ‘English’ response — a non-English language or a blank answer — produces the fail-closed ‘screen required’ fact. The survey never identifies the student on its own; it only states that the screener must be administered.
Step 2 · Screen and identify — a recommendation the committee acts on
The language-screener score is compared with the state education agency’s proficiency threshold to produce an identification RECOMMENDATION: identify, not-EL, or inconclusive-needs-committee. It is a recommendation only — every row carries the committee-determination guard, and the committee makes the determination under the SEA’s adopted criteria.
Step 3 · Serve and monitor — an active EL is owed services
For an active English learner, the owed-services answer is true and stays true — language services are never silently dropped. A reclassified student enters the post-reclassification monitoring window and is tracked for the required years. Status is resolved over an opaque student reference, never a name.
Step 4 · Reclassify — exit only when the criteria are met
When the committee weighs exit, the reclassification recommendation reuses the canonical exit gate: it recommends exit only for a current active EL whose supplemental criteria are all met, returns partial evidence to the committee, and never recommends a premature exit. The determination remains the committee’s; Linguistic supplies the fail-closed recommendation and its provenance.
The EL engines
Every feature is labelled honestly: Built means the underlying policy is pure and production-ready. There is no AI in any of them, and no live SIS sync yet — the surface computes over the roster you bring, with the consent gate applied against your tenant’s real consent records.
Title III intake starts with the home-language survey. Linguistic turns the three survey answers into a single fail-closed FACT: any answer that is not an explicit ‘English’ response — a non-English language, or a blank or unknown answer — triggers a required language screening. It never skips a screen on an ambiguous answer, and it never identifies the student from the survey alone. This is the screen-before-services trigger, and it is built and pure.
Built · fail-closed intake
A language-screener score plus the state education agency’s proficiency threshold becomes a committee-facing RECOMMENDATION: a score below the threshold recommends identifying the student as an English learner; a score at or above recommends not-EL (screened-out); a missing score or unusable threshold is inconclusive and returned to the committee rather than guessed. Every outcome carries a committee-determination guard — it is a recommendation, never an auto-final classification. The identification engine is built and pure, with no AI in the path.
Built · recommendation only
For each program record the status resolver returns the fail-closed Title III state: not-EL, identified-pending-screener, active EL, reclassified fluent-English-proficient, or in monitoring. The owed-services answer is true exactly when the student is an active EL — the never-drop headline: language services are never silently removed from an active English learner. The status and owed-services engine is built and pure; it reads an opaque student reference and never a name.
Built · never-drop headline
Exit (reclassification) is a RECOMMENDATION built directly on the canonical exit-criteria primitive: it recommends exit only when the student is genuinely a current active EL and the committee’s supplemental exit criteria — language-proficiency assessment, academic performance, and teacher evaluation — are all met; partial evidence is returned to the committee; no evidence is not-yet. A premature exit denies a student services they still need, so the gate is fail-closed and never re-implements the exit criteria — it reuses the canonical one and carries its reason as provenance. Built and pure.
Built · no premature exit
Linguistic never auto-decides a student’s classification or exit. Every identification and reclassification output object carries an explicit requires-committee-determination guard so a downstream surface can never mistake a recommendation for a final, machine-made legal decision. There is no AI anywhere in the path — the math is a deterministic threshold comparison and a reuse of the canonical exit gate. It is decision support for a human committee, and it is not legal advice: the criteria vary by state education agency and every production determination is the committee’s under the SEA’s adopted criteria.
Built · human decides, no AI
An EL subject’s name together with their home language reveal national-origin and language-minority status — sensitive student PII. Every emit of that PII routes through one consent chokepoint: a single tenant-scoped consent read per page (never one per student) applies the canonical directory-info basis, which is default-deny, and the do-not-publish kill-switch short-circuits every purpose. A subject whose consent is missing, lapsed, unresolvable, or publication-suppressed is redacted fail-closed — the name and home language are stripped while the row stays workflow-addressable. Minor student data is never made public and is never sold or shared. The consent gate is built and pure.
Built · consent-gated, tenant-scoped
For coordinators
An ambiguous or blank home-language answer triggers a screen rather than skipping one; a missing screener score is inconclusive rather than a guess; an active EL is always owed services; and a reclassification is blocked unless the criteria are met. The default is always to protect, plan, and screen — never to silently drop a student’s services or exit them early.
Identification and reclassification are recommendations, never auto-final classifications. Every such answer carries a committee-determination guard, there is no AI in the path, and the criteria are the state education agency’s. Linguistic is decision support and provenance for the committee — it is not legal advice.
An EL subject’s name and home language reveal national-origin, so every emit routes through one tenant-scoped consent read: default-deny directory-info consent plus a do-not-publish kill-switch. A subject without consent on file is redacted fail-closed — the name and home language are stripped while the row stays addressable. Minor student data is never public and is never sold or shared.
To book a conversation: [email protected].
What is built and what is coming — plainly
Built and pure today: the home-language-survey screen-before-services fact; the identification recommendation from the screener score and the SEA threshold; the program-status and owed-services resolver (an active EL is never silently dropped); the reclassification recommendation built on the canonical exit gate (never a premature exit); the post-reclassification monitoring-window fact; and the consent chokepoint over EL-subject PII (default-deny directory-info plus a do-not-publish kill-switch, fail-closed redaction, one tenant-scoped consent read per page). There is no AI in any of them.
Honest-off today: there is no persisted EL store and no live student-information-system sync. The surface is a Title III worksheet that computes over the EL roster you bring from your own export, with the consent gate applied against your tenant’s real consent records. Money is honest-off — there is no live checkout and no card is charged today. No billing. No subscription. No school has run a live EL program on linguistic.software yet. A conversation is the honest next step.
Connected to the platform
Linguistic is the Title III EL program-management layer: the screen-before-services fact, the identification and reclassification recommendations, the owed-services and monitoring answers, and the consent chokepoint over EL-subject PII. homeroom.software is the school operating system it belongs to — the roster, the records, and the consent system the EL surface reads its tenant-scoped consent from.
Early access · EL coordinators, advisers, and administrators
Linguistic is in active development. No school has run a live EL program on linguistic.software yet. We do conversations that show the current state honestly: how the home-language survey produces the fail-closed screen-required fact, how the screener score and the SEA threshold produce a committee identification recommendation, how the reclassification recommendation reuses the canonical exit gate, and how the consent gate redacts an EL subject’s name and home language fail-closed. There is no pricing commitment and no signup. If it looks right for your EL program, we discuss what early access looks like.
To book: email [email protected].
FAQ
No. Linguistic never auto-decides a student’s classification or exit. Identification and reclassification are RECOMMENDATIONS a qualified committee acts on — every such output carries an explicit requires-committee-determination guard. It is decision support, not a decision, and it is not legal advice. The committee makes each determination under the state education agency’s adopted criteria.
No. There is no AI anywhere in the path — no model, no inference, no generation. Identification is a deterministic comparison of the screener score against the SEA threshold, and reclassification reuses the canonical exit-criteria primitive. The same inputs always produce the same recommendation, and every recommendation is a plain, testable, deterministic result the committee can audit.
Intake begins with the home-language survey: any answer that is not an explicit ‘English’ response — including a blank or unknown answer — triggers a required screening, fail-closed. The language screener is then administered, and its score is compared with the SEA’s proficiency threshold: below the threshold recommends identifying the student as an EL, at or above recommends not-EL, and a missing score is inconclusive and returned to the committee rather than guessed.
Exit is a recommendation built on the canonical exit gate. It recommends exit only when the student is a current active EL and the committee’s supplemental criteria — language-proficiency assessment, academic performance, and teacher evaluation — are all met. Partial evidence goes back to the committee; no evidence is not-yet. A premature exit denies a student services they still need, so the gate is fail-closed. A reclassified student then enters the required post-reclassification monitoring window.
An EL subject’s name together with their home language reveal national-origin and language-minority status, so they are treated as sensitive student PII. Every emit routes through one consent chokepoint: a single tenant-scoped consent read applies the canonical directory-info basis (default-deny), and a do-not-publish kill-switch short-circuits every purpose. A subject with missing, lapsed, unresolvable, or publication-suppressed consent is redacted fail-closed — the name and home language are stripped while the row stays addressable. Minor student data is never made public and is never sold or shared.
No. Title III identification and reclassification criteria vary by state education agency, and every production determination is made by the qualified committee under the SEA’s adopted criteria. Linguistic supplies the fail-closed screen-before-services fact and the committee-facing recommendation with its provenance; it does not substitute for the SEA’s criteria and it is not legal advice.
Not yet. The engines — the screen-before-services fact, the identification and reclassification recommendations, the status and owed-services resolver, the monitoring-window fact, and the consent gate — are built and pure. There is no persisted EL store and no live SIS sync in the product today: the surface is an honest worksheet that computes over the EL roster you bring from your own SIS export, with the consent gate applied against your tenant’s real consent records. If you want to be an early school on linguistic.software, email [email protected].
Linguistic is the Title III EL policy layer and the committee workflow around it — the fail-closed screen-before-services fact, the identification and reclassification recommendations, the owed-services and monitoring answers, and the consent chokepoint over EL-subject PII — not a whole student information system. It is the EL program-management surface your coordinators run identification, service planning, and reclassification through, and it links back to homeroom.software, the school operating system.
There is no live checkout and no card is charged today — money is honest-off. There is no contract required and no signup. If the EL program-management worksheet and the committee workflow look right for your program, we discuss what early access looks like. Email [email protected] to book a conversation.