How to Use Speaking Stats to Improve Language Classes

After a class ends, most language teachers are left with a question that goes unanswered: did my student speak enough today? Is their fluency actually improving, or are we repeating the same stumbling blocks week after week?

Answering that from memory alone is hard. The class happens, it's over, and the following week you start fresh with no concrete data on what changed.

Why speaking performance is hard to track

In conversation classes, progress doesn't follow a straight line. One student can improve their pronunciation in March and slip on vocabulary in April. Another fills plenty of speaking time but cycles through the same 50 words. A third has solid vocabulary but speaks too slowly and stumbles at natural connection points.

Each of those profiles calls for a different approach. And identifying which pattern belongs to which student, class by class, is nearly impossible when you're taking notes by hand mid-conversation. You're running the class, not monitoring metrics.

The result is that most teachers make pedagogical decisions by gut feel. Sound gut feel, built from experience, but still without real data from the conversation itself.

What teachers typically do to measure student performance

The most common solution is asking the student: "did you feel you spoke more today?" The problem is that students' perception of their own performance is notoriously unreliable. An anxious student thinks they barely spoke even when they drove the whole conversation. A confident one thinks it went great even when they repeated the same mistakes throughout.

Another approach is jotting down observations mid-class, in a notebook or a doc on a second screen. It works to a point, but it breaks the flow of conversation and only captures what you managed to write down in the heat of the moment.

Some teachers record classes and watch them back afterward. That gives better data, but it eats up time most teachers don't have, and it still means scrubbing through the recording minute by minute to find the relevant moments.

None of these approaches scales when you have 15 or 20 students with different profiles.

What a good speaking stats system needs to deliver

To be genuinely useful, speaking performance metrics need to:

• Show the actual time the student spent speaking during the class, separated from the time you spent speaking.
• Report speaking pace in words per minute, since fluency is directly tied to rhythm.
• List the vocabulary that came up in that class, especially new words relative to the student's history.
• Be generated automatically, without any extra work on your part after the class.
• Be available to the student as well, so they can track their own progress.

Cross-referencing those metrics over time is what surfaces real patterns: the student who speaks fast but has a narrow vocabulary, the one who has built up a wide vocabulary but still hesitates constantly, the one who made a clear fluency leap over the past four weeks.

How Noladi calculates and displays speaking stats

After each live class ends, Noladi processes the recording and automatically generates conversation statistics for each participant.

In the lesson review, you and your student can see speaking pace in words per minute, total words spoken, the number of unique words used, and each person's speaking time. The split between your speaking time and the student's appears clearly, which helps you calibrate the balance for upcoming classes, especially in conversation-focused sessions where the goal is to get the student producing more.

Noladi also lists the new words that came up in that class based on the student's history. Over the weeks, you can watch the accumulated vocabulary grow, which is one of the most concrete ways to show a student that they are genuinely making progress.

AI correction suggestions appear alongside the statistics and flag grammar, pronunciation, and vocabulary points that stood out in that class. This adds pedagogical context to the numbers: it's not just that the student spoke less, but where and why they got stuck.

All of this is stored per class and accessible to both you and your student, with no manual work on your part once the session ends.

Get to know Noladi

If you want to track your language students' real performance using concrete data from each class, Noladi generates these statistics automatically after every live session. You and your student both have access to speaking pace, vocabulary, and participation time for every class, all in one place.

Learn more at noladi.app/teacher.