Teachers can add speaking practice at scale by moving some of it out of whole-class time and into short tasks that give each student feedback.
Most English teachers already know plenty of speaking activities. What they lack is the time to hear every student and respond.
Illustrative scenario. A junior high English teacher has five classes of 30 students, so 150 students in all. Hearing each student speak for one minute, once a week, takes 150 minutes, or 2.5 hours, before any feedback is given. These numbers are assumptions, not measurements, and the same teacher appears in every section below.
This guide is for principals, heads of English and district leaders who plan English programs. It covers where teacher time goes, what research says about speaking practice, how ELSA Schools fits beside a curriculum, and what each level of a school system can see.
What is teacher capacity in an English class?
Teacher capacity is the number of student responses a teacher can hear, assess and respond to in a week.
That's a working definition for this guide, not an established research term. It depends on hours, and a teacher's hours are fixed. The table shows how the week divides on average for lower-secondary teachers.
| Measure | Hours a week | Source |
|---|---|---|
| Total working hours, OECD average | 41 | Straits Times, 2025 |
| Classroom teaching, OECD average | 22.7 | Straits Times, 2025 |
| All other work (calculated: 41 minus 22.7) | 18.3 | Calculation |
| Total working hours, Japan | About 55 | OECD, 2025 |
| Total working hours, Singapore | 47.3 | Straits Times, 2025 |
Source: OECD Teaching and Learning International Survey (TALIS) 2024, as reported by the OECD and the Straits Times. Other work includes lesson planning, marking and administration.
What limits K-12 English teachers' time and attention?
Workload is the biggest limit: the OECD's TALIS 2024 survey names it the greatest demand on teachers.
The OECD survey collected answers from educators in 55 education systems. Japan reported around 55 hours a week.
Hours outside the classroom compete with listening
On the OECD averages above, about 18 of 41 weekly hours fall outside classroom teaching. In the scenario, hearing 150 students for one minute each would take 2.5 of those hours, every week.
A spoken answer leaves nothing to mark
A reading or writing task leaves a page a teacher can take home and mark. A spoken answer disappears unless something captures it. That makes speaking easy to skip when time runs short. This describes how the tasks work. It isn't a measured finding.
How much time do students actually spend speaking in English class?
The recent classroom studies reviewed here report talk for the whole class, not for each student.
One 2024 observation study of senior high English classes in Indonesia found that students produced 45.6% of verbal behavior in the first meeting observed, and the teacher about 49% (Supriadin et al., 2024). The study is descriptive and covers one school.
A class-level share hides the per-student share.
Illustrative scenario. Suppose students speak for half of a 45-minute lesson, which is 22.5 minutes. Shared across 30 students, that is 45 seconds each, and only if every student gets an equal turn. An earlier ELSA guide on AI speech analysis works through a 40-student case with different assumptions.
Does more speaking practice improve proficiency?
More speaking practice is linked to better fluency, but the studies reviewed here mostly involve older learners.
A 2025 study of two Bangladeshi university classes compared 30 hours of interactive speaking practice with 20 hours of monologue tasks. The first class improved fluency significantly more (Haque et al., 2025). The classes differed in both hours and activity type, so the study can't show how much of the gain came from the extra time.
For AI-supported practice, a 2025 meta-analysis of experimental and quasi-experimental studies reports a large overall effect on English speaking and listening skills, with a standardized mean difference of 0.98 and a 95% confidence interval of 0.57 to 1.39 (Jantakoon et al., 2025). A 2025 paper in the Australasian Journal of Educational Technology points to methodological limits in this body of research.
None of the sources reviewed here measures the link between speaking practice and proficiency for K-12 students specifically. A practical reading is to treat extra speaking practice as a reasonable bet and to measure it in your own classes.
How does ELSA Schools add speaking practice without more teacher work?
ELSA Schools turns a teacher's existing materials into speaking assignments, so students practice and get feedback without the teacher listening to each attempt.
Teachers upload vocabulary lists, textbook passages or reading materials, or ask AI to draft new content. ELSA Schools converts them into speaking activities and assigns them to a class in one step. Students practice on their own in a web browser, and the teacher reviews results in a dashboard.
| Practice type | What the student does | What is scored | How the teacher sees it |
|---|---|---|---|
| Vocabulary and sentences | Says set words or sentences | Pronunciation and delivery | Class dashboard |
| Reading aloud | Reads a full passage | Pronunciation, intonation, fluency | Class dashboard |
| Real conversations | Talks with an AI partner using lesson vocabulary | Clarity, pronunciation, delivery | Class dashboard |
| Speech Analyzer | Records own words on any topic | Pronunciation, intonation, fluency, vocabulary, grammar | PDF report the student shares |
One school leader described the benefit as less time spent on assessments and more effective practice (Hoang Mai Star School story). That's one leader's view, not a measured time saving.
What does a weekly workflow look like beside the curriculum?
A weekly cycle adds one assignment, one in-lesson practice slot and one dashboard review to the lessons a teacher already plans.
The scenario teacher follows a textbook unit on daily routines, with two lessons a week.
Set up the unit
Before the unit, the teacher uploads the unit's word list and reading passage. ELSA Schools turns them into activities. The textbook still sets the order and the content.
Assign practice before the lesson
The teacher assigns a study set and a read-aloud task for home. Shibuya City's junior high schools used read-aloud practice at home this way, and teachers checked progress on the dashboard before the lesson.
Use five to ten minutes in the lesson
A short slot at the start of the lesson lets every student practice at the same time. The teacher walks around and looks at screens. Some Japanese schools run this as a 5 to 10 minute block.
Add one AI role-play to the main activity
Shibuya added AI role-play to its existing language activities as a new option and didn't replace them.
Review the dashboard at the end of the week
The teacher checks assignment completion and the class's weakest skill, then picks a few students for a short conversation. In the scenario, talking with 10 students for one minute each takes 10 minutes, against 150 minutes for the whole group. This is an assumption.
| If you see | Do | Owner |
|---|---|---|
| Several students haven't finished the assignment | Fix an in-lesson slot, or reassign for the next lesson | Teacher |
| One skill is low across the class | Reteach it in the next lesson | Teacher |
| One student scores low on specific words | Hold a short live conversation about those words | Teacher |
| Completion drops across several classes | Raise it at the department meeting | Head of English |
| Participation is low across the school | Review lesson slots and device access | Principal |
How does speaking practice become visible to principals and district leaders?
One dashboard shows each class and student, and reports can be shared with parents and school leaders.
Each level of a school system needs a different view. The table shows who looks at what, and a suggested rhythm.
| Level | What they look at | Suggested rhythm | Decision it supports |
|---|---|---|---|
| Student | Own scores and feedback after each attempt | Every attempt | What to practice next |
| Teacher | Individual performance, class insights, assignment completion | Weekly | Who to talk to, what to reteach |
| Head of English | Completion and skills across classes | Monthly | Where to change lesson slots or training |
| Principal | Progress reports for school leadership | Each term | Whether to extend to more classes |
| Parents | View-only dashboard access | Each term | How to support practice at home |
| District or ministry | Results shared across schools | Each term or year | Training, contests and rollout pace |
Kyotango City's Board of Education shows the district level. It began with 100 volunteer students in 2021 and by 2024 had extended ELSA Schools to all fifth and sixth graders and all junior high students. It ran four teacher training sessions a year and shared results and issues across the city's teachers.
What do real rollouts show?
Reported results from Shibuya, Kyotango and Hoang Mai are encouraging, but each reflects one setting and not a guaranteed outcome.
| Setting | Scale | What was reported | Scope note |
|---|---|---|---|
| Shibuya City Board of Education, Tokyo | All 8 public junior high schools | Speaking test (ESAT-J) scores up 10.4 points from 2023 to 2025; share of students at CEFR A1 or higher up 9.5 points; between April and October, students who devise their own ways to learn English rose from 67% to 85%, and students who say they converse well with classmates rose from 47% to 72% | Lessons also involve teachers and assistant language teachers, so gains aren't ELSA alone. Survey items are self-reported |
| Kyotango City Board of Education, Kyoto Prefecture | From 100 students in 2021 to all grade 5 to 6 and junior high students in 2024 | About 80% of students said their English improved | Self-reported survey |
| Hoang Mai Star School, Vietnam | 1,000 students | 79% showed measurable improvement in English proficiency; 74% weekly engagement; 2,500 practice hours in three months | One school. Figures as reported in its customer story |
Each setting measured different things over different periods. Read them as examples of how rollouts run, not as forecasts.
Where does this approach fall short?
It adds practice and visibility, but it doesn't replace teacher-led speaking, and K-12 evidence for the link to proficiency is still limited.
Limits to know
- Scores guide practice. ELSA Schools scores are a guide for improvement, not a grade.
- Devices and time. Students need a device with a browser and a protected slot in the lesson.
- Teacher judgment. Teachers still decide what to teach and which students need a conversation.
Common mistakes
- Replacing teacher-led speaking activities. Add practice to them.
- Treating the score as a grade. Use it to decide who needs a conversation.
- Assigning practice without a lesson slot. Protect a short slot so every student practices.
- Reading only completion. The skill breakdown shows what to reteach.
- Starting with every class. Pilot a few classes and review the dashboard for a term first. After a demo, ELSA offers a 14-day trial with 5 to 10 accounts, and the minimum package is 50 student licenses on a 12-month term.
Related reading
- ELSA Schools: AI speaking practice for K-12 classrooms
- AI speech analysis for schools: how it helps students
- Blended learning: boosting student confidence and engagement
- Hoang Mai Star School customer story
Frequently asked questions
How much of a teacher's week goes to teaching?
On OECD averages for lower-secondary teachers in TALIS 2024, about 22.7 of 41 weekly working hours go to classroom teaching. The remaining 18.3 hours cover work such as lesson planning, marking and administration. Japan reported around 55 hours a week.
Can one teacher hear every student speak each week?
Rarely. In the scenario of five classes of 30 students, hearing one minute from each student takes 150 minutes, or 2.5 hours, with no time for feedback. Short assignments that give students feedback, plus a dashboard that shows who needs a conversation, reduce how much a teacher needs to hear.
Does AI speaking practice improve proficiency?
A 2025 meta-analysis reports a large overall effect of AI-enhanced instruction on English speaking and listening. A 2025 journal paper notes methodological limits in this research. None of the sources reviewed here measures the effect for K-12 students specifically, so schools should measure results in their own classes.
How can a principal tell whether students are practicing speaking?
ELSA Schools has one dashboard for every class and student. It shows assignment completion, individual performance and class-level insights, and reports can be shared with school leaders and parents. A principal can review completion and skill breakdowns each term and ask the head of English where practice is falling behind.
Does ELSA Schools replace the English teacher?
No. ELSA Schools adds practice alongside lessons. The teacher still chooses what to teach, runs the speaking activities in class and decides which students need a conversation. Shibuya City added AI role-play to its existing language activities as a new option and didn't replace them.
Can a school try ELSA Schools with a few classes first?
Yes. After a demo, ELSA offers a 14-day trial with 5 to 10 accounts. The minimum package is 50 student licenses on a 12-month term, and teacher and admin accounts are unlimited. A few classes is a sensible starting point before extending to the school.
Next step
- Teacher capacity is the number of student responses a teacher can hear and answer in a week.
- Whole-class talk time hides how little each student speaks, and the research on speaking practice is mixed for K-12.
- A weekly cycle of one assignment, one in-lesson slot and one dashboard review adds practice without adding a new lesson.
To see ELSA Schools with your own classes, book a demo to unlock a free trial.