Validity and authenticity of Auto-Detect

Question

We have tried using Reading Progress with automatic grading on for English read-aloud assignments for the following purposes.

-Eliminate discrepancies in grading criteria among multiple teachers

-To shorten the time required for grading oral reading assignments.

One thing we are not satisfied with is the grading.

Pronunciations that are not right in any way are passed over, while parts that do not interfere with communication at all are marked.

Some of the evaluations are unsatisfactory even from a native speaker's point of view.

How does the AI evaluation work?

I would like to know the details of what exactly is being evaluated, rather than abstract things like "fluency”. Without knowing what points to correct, the students will lose the motivation.

If its AI grading is refference only, teachers will have to grade by themselves anyway. Then what’s the point?

Answer 1

Hi @42799510

Thanks for visiting our forum. Microsoft Teams tag is mainly focused on the general issue of Microsoft Teams application troubleshooting. Based on your description, this problem is related to Teams Edu. The following is for informational purposes only:

The Auto-detect feature is a PREVIEW feature in Reading Progress that supports a number of dialects and pronunciations. It evaluates student recordings to identify likely mispronunciations and other reading errors for educators. It also calculates the accuracy rate, correct words per minute, and most challenging vocabulary for each student. You can adjust the pronunciation sensitivity of Auto-detect to make it more or less strict when listening and estimating errors. You can also edit the language of the uploaded document to ensure Auto-detect uses your intended language.

The Auto-detect feature is not perfect and may not recognize accents and dialects well. It may also miss some errors or mark some correct words as errors. Therefore, it is recommended that you review the student recordings and manually mark any errors that Auto-detect did not catch or correct any errors that Auto-detect wrongly marked. You can also turn off the Auto-detect feature if you prefer to mark all the errors yourself.

The validity and authenticity of Teams Auto-detect depends on several factors, such as the quality of the student recording, the clarity of the student speech, the language and dialect of the student, and the pronunciation sensitivity setting. It is important to note that Auto-detect is not a substitute for human evaluation, but rather a tool to assist you in providing personalized feedback and instruction for your students. You can use the data collected by Auto-detect in Insights to track your students’ progress over time and identify areas of improvement.

---

If the answer is helpful, please click "Accept Answer" and kindly upvote it. If you have extra questions about this answer, please click "Comment".

Note: Please follow the steps in our documentation to enable e-mail notifications if you want to receive the related email notification for this thread.

---