A Novel Framework for Augmenting Rating Scale Tests with LLM-Scored Text Data