How does MAP Reading Fluency account for students with accents?

Document created by Community User on May 3, 2018Last modified by Community User on May 4, 2018
Version 2Show Document
  • View in full screen mode
How does MAP Reading Fluency account for students with accents?


How does MAP Reading Fluency accurately score students with accents?

The underlying AI-engine used in MAP Reading Fluency includes speech scoring technology developed by SRI International, which is the same company that developed Apple's Siri. This technology is extremely robust and was originally developed to evaluate the understandability of non-native English speakers. MAP Reading Fluency uses a simplified pass-or-fail model at the word level to generate a words-correct-per-minute score. We have set the pass or fail thresholds to achieve the best match between human judges and the AI-engine. We use a separate value that denotes the "confidence level" of the AI score, and we exclude any record that does not pass this threshold from scoring. For more information, see How is audio rejected for MAP Reading Fluency? 

Among the remaining records, the inter-rater agreement between human experts and the AI values is up to .99 in our sample (where 1.0 is perfect agreement.)

See also: Can MAP Reading Fluency be used for English Language Learner (ELL) students?

What happens if MAP Reading Fluency cannot score a student's assessment?

For any unscored record, the teacher has the option to manually score the child's reading using the audio playback and in-app keystrokes. In the case of a student with an accent, familiarity with the student may help the teacher judge correctness when the actual audio signal is ambiguous. See How to perform self-scoring for audio recordings in MAP Reading Fluency.

Article Number