Start with the scale, because it is larger than most people assume. In 2025, 40.0% of NSW public school students — 312,242 children — came from a home where a language other than English was spoken, up from 32.3% just a decade earlier [EAL-01]. That is not simply a reflection of Australia's general cultural mix: the 2021 Census found 26.4% of the whole NSW population spoke a language other than English at home, so the school-age population is running well ahead of the general population in linguistic diversity, driven by younger, more recently arrived families entering the public system [EAL-01]. Nationally, the 2021 Census recorded 5.6 million people — 22% of all Australians, all ages — using a language other than English at home, up from 20.6% in 2016, with 3.4% of the whole population reporting they spoke English 'not well or not at all' [EAL-02]. Two in five NSW public school students, then, come from homes where a language other than English is spoken by the student and/or a parent or carer. That is the ordinary condition of a modern Australian classroom, not an exception to be managed.
It is also, critically, not one condition. NSW's own Department of Education is explicit that Language Background Other Than English (LBOTE) and English as an Additional Language or Dialect (EAL/D) are different things, and warns against collapsing them: LBOTE is a broad demographic category — anyone from a home where another language is spoken — while EAL/D is a needs-based educational category describing students currently learning English and requiring support to access the curriculum [EAL-03]. 'Not all LBOTE students are EAL/D,' the Department states plainly, 'as many may be fluent (proficient) in English' [EAL-03]. The numbers bear this out: of the 40.0% of NSW public students who were LBOTE in 2025, an estimated 65.4% were also classified as EAL/D learners — 26.6% of all students, or 205,555 children [EAL-03]. The remaining third of LBOTE students were not currently flagged as needing English-language support at all. Confusing the demographic label with the learning need is the single most common error in how this cohort gets discussed, and the government's own fact sheet corrects it directly.
That distinction matters because the achievement data tells a story of range, not deficit. ACARA's 2022 NAPLAN National Report shows LBOTE students scoring equal to or higher than non-LBOTE students on average across nearly every domain and year level tested — reading, writing and numeracy, Years 3, 5, 7 and 9 — a pattern holding consistently back to 2008 [EAL-07]. In Year 3 reading, the LBOTE mean was 447.8 against 434.2 for non-LBOTE students; in Year 5 numeracy, 500.7 against 483.5; in Year 9 numeracy, 597.5 against 580.3 [EAL-07]. But the mean is only half the picture, and it is the half that flatters a tidy story. In every one of those same cells, the LBOTE standard deviation runs noticeably larger than the non-LBOTE figure — in Year 9 numeracy, 70.2 against 58.4 [EAL-07]. A higher average paired with a wider spread means the LBOTE cohort contains more students at the very top and more students who are genuinely struggling, both at once, than the non-LBOTE comparison group. Averaging over that range and reporting only the mean would hide the very students most likely to need help — which is exactly the trap the LBOTE/EAL/D distinction exists to prevent.
That range shows up again at the point of formal assessment. In 2024, VCE English as an Additional Language had 2,583 total enrolments against 45,751 for mainstream English — roughly one-eighteenth the size, confirming EAL is a genuine minority pathway within the senior secondary system, not a marginal footnote [VCE-15]. On the written examination, EAL students averaged 31 out of 60 (median C+, std dev 9.2) across 2,276 assessed candidates [VCEL-01]. The 2024 examiners' report — the first sitting of a new study design — found the lowest-scoring responses shared a specific, nameable set of habits: focusing on a single word or idea in the essay topic rather than the whole of it, misinterpreting or disregarding the set topic, leaning heavily on plot summary as evidence, and writing in the register of a review rather than an analytical response [VCEL-03]. None of that describes a language deficit in the conversational sense — it describes the specific, teachable skills of academic argument: reading a prompt precisely, building an analytical rather than a narrative structure, using evidence as evidence rather than as plot recap.
Why would a wide gap open at exactly the essay-writing stage while the broader NAPLAN averages run level or ahead? The research on second-language acquisition gives a consistent, decades-old answer: conversational fluency and academic language proficiency are not the same skill and do not develop on the same timeline. Jim Cummins' widely used BICS/CALP distinction holds that conversational fluency typically reaches a functional, peer-appropriate level within about two years of exposure, while the academic aspects of a language take five to seven years, on average, for immigrant students to approach grade-level norms [EAL-06]. A landmark Stanford study of English learners across four California and Canadian school districts — including the two California districts judged most successful at teaching English to limited-English-proficient students — found oral proficiency developing in three to five years, but academic English proficiency taking four to seven [EAL-05]. This is not a fringe or foreign finding: it is built directly into Australian policy. NSW's own EAL/D advice for schools gives near-identical official timeframes — four to seven years to develop academic language for students with prior first-language education, up to ten years for students with disrupted first-language schooling, and seven to twelve years for students with limited education in any language [EAL-04]. VCAA's own VCE eligibility rules encode the same reality: a student may only sit EAL rather than mainstream English if they have not been taught in English, or resided in Australia, for more than seven years — VCAA's own recognition that seven years is roughly the point at which academic English catches up to a first-language peer [EAL-08]. A student can sound completely fluent in the playground within two years and still be years away from the sentence-level command an analytical essay demands — not because they are behind, but because that is how long the specific register being tested takes to build for everyone.
That distinction has taken on a new, sharper edge with the rise of AI writing detectors in schools. A 2023 peer-reviewed study testing seven widely used GPT-detection tools found they achieved near-perfect accuracy on native-English (US eighth-grade) essays, but misclassified more than half of non-native (TOEFL) essays as AI-generated — an average false positive rate of 61.22% [AIE-05]. Simple rewriting cut that rate to 11.77%, showing the bias was measurable and largely an artefact of vocabulary predictability rather than any real signal of authorship [AIE-05]. For Australian EAL/D students and international students, many of whom are already working harder to produce the same academic register their peers reach faster, a teacher's AI detector can penalise the very evidence of effort — plainer, more formulaic sentence construction — as if it were a shortcut. It is a modern, automated version of the same old mistake: judging a still-developing academic register as if it were a deficiency rather than a stage.
Put together, the evidence resists a single tidy headline, and that is the honest finding. Two in five NSW public students speak another language at home; most of them are not currently flagged as needing English support at all, and among those who are, achievement data shows higher averages paired with wider spreads — real strength and real struggle coexisting inside the one LBOTE label, not a uniform story either way [EAL-01] [EAL-03] [EAL-07]. What the evidence converges on is not a claim about ability but a claim about time and register. Conversational English arrives in a year or two; the academic English of analysis, structured argument and the formal essay takes four to seven years or more, is written into Australian curriculum policy and VCE eligibility rules alike, and is precisely the register examiners flag as missing in the lowest-scoring EAL responses [EAL-06] [EAL-04] [EAL-08] [VCEL-03]. That is the specific, buildable skill — not fluency, which mostly looks after itself, but the deliberate architecture of an analytical paragraph — that sustained one-to-one writing instruction is built to close.
