Reading the Science Test Aloud: Accommodations or a Different Construct?
Riverside Middle School has nine ELs in Mr. Okafor's sixth-grade science class of 28. Their WIDA levels range from 1.5 to 3.2. The upcoming ecosystems test has 22 multiple-choice items, four short-answer items, and stimulus passages of about 120 words each, written at roughly a seventh-grade reading level. Mr. Okafor proposes reading the entire test aloud to the ELs in a separate room and allowing word-to-word bilingual dictionaries. Two weeks ago these same students scored well on a hands-on food-web lab practical. One student, Yeni, arrived from Guatemala in November after three years of interrupted schooling; she is at WIDA 1.5, reads at about a first-grade level in Spanish, and produces single-word answers in English. The principal wants a decision by Friday so the test can be copied and a room reserved, and she has asked you to state exactly what will happen for Yeni.
1. Evaluate Mr. Okafor's two proposals. For each, say whether you would keep, modify, or reject it, and explain your reasoning about what this test is supposed to measure.
Read the worked answer
Start from the construct. This is a science test on ecosystems, not a reading test, so removing unnecessary reading barriers is legitimate: it lets students demonstrate the knowledge being graded. Read-aloud should be kept for students at roughly WIDA 1.5–2.5 who receive oral support daily, because an accommodation must mirror what the student uses in instruction or it is not valid. Reading the stimulus passages aloud does not corrupt a science construct, but it does mean the teacher must decide in advance that no item is being scored as reading comprehension. Bilingual dictionaries should be modified, not rejected. A word-to-word dictionary helps the students at 2.5–3.2 who are literate in Spanish or Arabic; it does nothing for Yeni, whose Spanish literacy is at a first-grade level, and it adds cognitive load. For her, replace it with a visual word bank and glossed key terms. Both proposals need a third element Mr. Okafor did not mention: extended time and a separate setting, so read-aloud does not stigmatize students in front of peers. Do not reduce the number of items; reduce linguistic complexity by shortening stems, bolding key terms, and adding diagrams. Finally, record every accommodation in the students' files so the pattern stays consistent across classrooms and matches what will be allowed on state assessments.
2. What evidence in the scenario should drive the decision for Yeni specifically, and what would you recommend in place of the standard test for her?
Read the worked answer
Three facts matter most: she has been in an English-medium classroom for about three months, her L1 literacy is at a first-grade level, and she already demonstrated content mastery on the food-web lab practical. That last point is the key evidence. It shows her difficulty is with the linguistic demands of the test, not with the science, and it gives you a validated model for an alternative format.
A 22-item test with 120-word passages is not accessible to her at any accommodation level, because reading aloud still delivers English text she cannot yet process at that speed. Instead, build a performance assessment on the same standards: label and sequence a food web from pictures, sort organisms into roles, match vocabulary to images, and answer three or four oral questions with visual supports, scored on a content rubric that does not penalize grammar. If a Spanish-language interpreter or a translated version is available, offer the written portion in Spanish, since her Spanish literacy, though low, is stronger than her English.
What you should not do is exempt her. Exemption would mean no evidence of her science learning, no data to plan instruction, and a message that content expectations do not apply to her. Keep her accountable to the same standards through a different route, and note the alternate format in her file so it is a documented, temporary decision rather than an informal favor.
3. A colleague suggests exempting all nine ELs from the unit test and grading them on effort and participation instead. How do you respond?
Read the worked answer
Three problems, in ascending order of seriousness. First, instructional: without assessment data you cannot tell whether the students are learning ecosystems or only attending, and you will not catch the students who need intervention. Second, equity: effort grades hide learning gaps and can inflate a student's record, which then collapses when the student meets a graded, credit-bearing course. Third, legal and civil rights: ELs are entitled to meaningful access to the content curriculum and to be assessed on it, and a pattern of excluding them from assessment can support a finding that the program is not providing that access.
Offer a better route. Keep the same content standards and vary access: read-aloud, visual word banks, shortened stems, extended time, oral response options, and, for the newest arrival, a performance-based format. Use a two-part rubric that scores science content separately from language so a student's content grade is not drowned by English errors. Report ELD progress separately through the WIDA standards and can-do descriptors, which is where the language information actually belongs.
If the concern is that one test does not capture the students' learning, the answer is multiple measures over time, not exemption. Add lab notebooks, a short performance task, and a teacher observation checklist to the evidence set, and let those measures carry weight. That preserves accountability while giving the students a fair chance to show what they know.