- Calculated for the association between spoken/written thoughts and sentences containing distal cause to differentiate between participants of varying reading abilities 
- High cosines for current sentence and distal cause 

University of Michigan - Ann Arbor

Measuring discourse-level processes with verbal protocols and latent semantic analysis - results

Millis, K., Magliano, J., & Todaro, S. (2006). Measuring discourse-level processes with verbal protocols and latent semantic analysis. Scientific Studies of Reading, 10(3), 225-240. https://www.tandfonline.com/doi/abs/10.1207/s1532799xssr1003_2?casa_token=t1DCNTJm4G0AAAAA:4Zmsq1hksZB2rZsEYOs0VuIGfn16rDbUviE39Om4DyMxLz0HWOC7RzN9yyymRidOj9nS1VQqXTtWZA

Measuring discourse-level processes with verbal protocols and latent semantic analysis

- Short answer test scores averaged for number of correct answers over question type to get 2 scores, one for text-based and one for situation model
- Average short answer test scores were 2.41 for text-based and 1.49 for situation model questions
- Nelson-Denny test mean of 26.7 

Measuring discourse-level processes with verbal protocols and latent semantic analysis - results: test scores

Measuring discourse-level processes with verbal protocols and latent semantic analysis - results: LSA cosines

- 3 multiple regression models for 3 measures of reading comprehension – text-based, situation model, Nelson-Denny reading comprehension test
- Cosines averaged for each participant, with a dummy variable of text to account for different texts read  
- Statistically significant equations demonstrate that mean LSA cosines could be used to predict reading comprehension
- The more similar a participant’s thoughts were to the causal antecedents rather than the sentence itself, the better the reader comprehended the passage 
- Comparing reading comprehension predictions from LSA cosines and Nelson-Denny test, cosines provided better predictions (i.e. 14% for textbase and 16% for situation model using cosines compared to 3% and 6% from Nelson-Denny)

Learn Before

Related