In this study we investigated the importance of language in student test pe
rformance on mathematics word problems. Students were given released items
from the National Assessment of Educational Progress mathematics assessment
, along with parallel items that were modified to reduce their linguistic c
omplexity. In interviews, students typically preferred the raised items ove
r the original counterparts. Paper-and-pencil tests containing original and
revised items were administered to 1,174 8th grade students. Students who
were English language learners (ELLs) scored lower on the math test than pr
oficient speakers of English. There were also differences in math performan
ce with respect to socioeconomic status (SES) but not gender. Linguistic mo
dification of test items resulted in significant differences in math perfor
mance; scores on the linguistically modified version were slightly higher.
Some student groups benefited more from the linguistic modification of item
s - in particular, students in low-level and average math classes, but also
ELLs and low SES students.