A Spanish-Speaking Child and Ten Read-Aloud Items
Sofia (6 years 1 month) has been in an English-medium kindergarten for eight months and speaks Spanish at home. On ten read-aloud sentence-completion items she answers three correctly. On those three she whispers a Spanish word before pointing, and the examiner recognizes each whispered word as the correct concept. On six pictorial-reasoning items that require no English she answers all six correctly, most in under ten seconds. Her teacher describes her as 'quick with puzzles, quiet in circle time.' The examiner must submit scores today and is considering reading the ten verbal items again in Spanish, using the same answer key and the same scoring rules. The school's testing policy permits directions to be repeated but not translated, and the examiner is unsure whether a translation counts as a repetition.
1. Should the examiner translate the ten verbal items into Spanish and score the answers against the same key? Argue both sides and reach a decision.
Read the worked answer
Case for translating: the items are read aloud, so a child who does not yet understand spoken English is being scored on English rather than on the underlying reasoning; translating would let her show what she knows; the pictorial-reasoning result suggests the reasoning is there.
Case against, which wins:
- It is not a repetition. A repetition reproduces the same words. A translation changes the stimulus, so the policy's permission does not cover it.
- The norms do not apply. Standard scores come from children who heard the English version. A Spanish-read item may be easier or harder, and the resulting number cannot be compared to the norm group or to her classmates.
- The construct includes English comprehension. Verbal comprehension items measure understanding of the language of instruction - a legitimate thing for a school to want to know.
- Item difficulty shifts unpredictably in translation: cognates, article patterns, and length all change the load.
Decision: do not translate for a score. Score the standard administration as given, and describe the language confound explicitly in the report.
2. Interpret the split between 3 of 10 verbal items and 6 of 6 pictorial items. What explanations fit, and which can be ruled out?
Read the worked answer
Fits the data:
- English comprehension is still developing. Eight months of exposure is short, and the whispered Spanish words show she is retrieving the right concept in her stronger language and then searching for the English label - a classic sequential-bilingual pattern.
- The verbal score is confounded. It mixes reasoning with English vocabulary and cannot be read as a pure measure of her thinking.
- Nonverbal reasoning is a relative strength. Six of six, fast, with no language demand.
Ruled out or unsupported:
- Low general ability - inconsistent with the perfect nonverbal performance.
- Not trying / attention - she answered all sixteen items and whispered translations, which is engagement, not avoidance.
- A hearing problem - she responded correctly to spoken English on the three items she got, and to spoken directions throughout.
What the data cannot do is quantify how much of the verbal gap is language and how much is anything else.
3. Write the recommendation section of Sofia's report. What should the school do with these scores, and what must the report avoid?
Read the worked answer
Recommended wording:
'Sofia completed the verbal items in English as standardized. She answered 3 of 10 correctly; on each correct item she first produced the concept in Spanish, indicating that the difficulty lies in English vocabulary rather than in the underlying knowledge. She answered 6 of 6 nonverbal pictorial-reasoning items correctly and quickly. The verbal score should be interpreted as a measure of English comprehension at eight months of exposure, not as an estimate of her reasoning ability. Placement and instructional decisions should rest primarily on the nonverbal results and on classroom observation. Re-assessment is recommended after a further period of English exposure; a bilingual evaluation would give a fairer picture of her verbal skills in the meantime.'
Must avoid:
- Reporting the verbal score as a general ability estimate, or averaging it with the nonverbal score into a single number.
- Any label - language delay, learning disability, low ability - from sixteen items.
- Implying the test was administered incorrectly, when the standard administration was in fact followed.