Abstract
<p>Cultural representation in K-12 STEM curriculum shapes whether students from underrepresented backgrounds see themselves as belonging in science. Hammond's neurological framework for culturally responsive teaching argues that students whose cultures are absent from learning materials face measurable cognitive consequences, yet no scalable method exists for quantifying cultural representation in curriculum text. This study introduces the Cultural Representation Index (CRI), a natural language processing tool built in Python using spaCy that scores curriculum texts on six weighted dimensions: diverse scientists, diverse locations, cultural concepts, funds of knowledge, multilingual markers, and identity representation. The CRI was applied to 44 K-12 and college-preparatory STEM texts drawn from CK-12 and OpenStax, validated against human ratings on a 20-text subset (95% observed agreement, Cohen's Kappa = 0.64, Substantial). Every text in the corpus scored below 40 out of 100, placing the entire sample in the Low Representation tier. The corpus mean was 9.99. OpenStax texts averaged 16.59 compared to 3.39 for CK-12, a 4.9-fold gap that persisted across both math and science. The diverse scientists dimension averaged 1.22, the lowest of any dimension, with 38 of 44 texts naming no scientist from an underrepresented background. These findings demonstrate that cultural representation in STEM curriculum is measurable, that the gaps are systematic rather than incidental, and that the platform serving the most under-resourced students shows the lowest representation.</p>