Assessment Design for a Mixed-Proficiency Latin I Class
You teach one section of Latin I with 27 students: 9 have IEPs specifying decoding and processing-speed accommodations; 3 have documented anxiety that worsens on timed translation; 2 are heritage speakers of Spanish with strong morphological awareness but no Latin; the rest are typical beginners. The unit covered third-declension nouns, present-tense sum and possum, and simple subject-object sentences. Your department requires a 50-minute summative assessment scored on a 4-point proficiency rubric. A colleague proposes a 40-item multiple-choice vocabulary and translation test with a 30-minute time limit. You must decide what to keep, change, or drop so the assessment measures the unit's Latin objectives rather than test format.
1. Identify the main validity threats in the colleague's proposal.
Read the worked answer
The proposal confuses speed and recognition with the unit's actual objectives. A 40-item multiple-choice test in 30 minutes measures rapid test-taking and decoding fluency more than the ability to read simple Latin sentences. For the 9 students with decoding and processing-speed IEPs, the time limit creates construct-irrelevant variance: their Latin knowledge may be adequate while their access is not. The 3 students with anxiety may similarly underperform. Multiple-choice vocabulary items can test isolated glosses rather than morphological parsing or sentence-level meaning, so the assessment may underrepresent the unit's syntax and reading objectives. The 4-point proficiency rubric also suggests performance tasks, not a raw MCQ score. Keeping some recognition items is fine, but the test must be aligned to third-declension forms, sum/possun, and simple sentence comprehension, with accommodations built in rather than added afterward.
2. Design a 50-minute assessment structure with sections and approximate timing, and justify each choice.
Read the worked answer
A defensible structure is: (1) 10 minutes-form and function table: students identify case/number and match third-declension forms to functions; (2) 15 minutes-sentence translation with a short glossed word bank, including sum and possum in context; (3) 15 minutes-a 40–50-word connected passage with 5 comprehension questions (main idea, reference, sequence, inference); (4) 10 minutes-one short constructed response asking students to explain a form or correct an error. This samples vocabulary, morphology, syntax, and reading without making speed the dominant construct. Students with extended-time IEPs receive the extra time in a separate setting; the 3 anxious students may take the passage section untimed or in a small group. The 4-point rubric should describe accuracy, comprehension, and use of evidence, not just number correct. The sections are short enough that decoding load does not swamp comprehension.
3. What evidence would you collect to show the assessment is fair across the class?
Read the worked answer
First, build a simple alignment table mapping each item or task to the unit objectives: third-declension case/number, sum/possum forms, and simple sentence comprehension. Items that do not map should be revised or dropped. Second, after scoring, disaggregate results by subgroup-IEP, anxiety accommodation, heritage Spanish-while keeping groups small enough to avoid labeling individual students. Look for patterns: if IEP students lose points mainly on timed items but perform on untimed parsing, the timing is a validity problem. Third, run a quick item analysis: items everyone misses may be poorly written or outside the unit; items everyone gets right may not discriminate. Fourth, ask two colleagues to review the language load and cultural references for construct-irrelevant difficulty. Finally, compare written responses to the rubric and keep anchor examples so scoring is consistent. Fairness is an evidence-gathering process, not a single accommodation.