Nobody Teaches Teachers How to Share a Classroom

Co-teaching in an elementary classroom

It is 9:40 on a Tuesday. The push-in ML teacher walks into a third grade classroom carrying a bag of sentence strips and a plan she made on her own the night before. The lead teacher is mid-lesson. He looks up, smiles, and says, “Oh good, you’re here. Can you sit with Miguel and Fatima?”

She sits with Miguel and Fatima.

Forty minutes later she walks to the next room and does it again. By June, she has spent an entire year as a highly trained specialist doing the job of a classroom aide. This is not because anyone decided that. Rather, it is a result of nobody deciding anything.

I have watched this scene play out in hundreds of schools over 30 years of training teachers. Nobody planned it this way. The lead teacher is overwhelmed, not hostile. The ML teacher is skilled, not passive. Both of them want the partnership to work, and both of them are already good at their jobs.

What is missing is not willingness. It is a framework, a shared lexicon, and some communication skills. Nobody teaches teachers how to share a classroom.

This post is a step in that direction.

Researchers have a name for the arrangement I just described: one teach, one assist. One teacher delivers the lesson and the other circulates, redirects, and helps a handful of students. It is the most common co-teaching model in American schools. It is also the weakest.

That is not an opinion. Andrea Honigsfeld and Maria Dove, whose research on co-teaching for English learners is the foundation of everything I am about to describe, have been clear on this for years. When classroom teachers and language teachers systematically co-plan and co-teach, multilingual learners gain. When one teacher teaches and the other assists, the specialist’s expertise never reaches the whole room, and the students who need her most get a whisper of it at the back table.

Districts fund push-in because it is supposed to be inclusive. On the surface, it sounds great. We think, “Well, there are two experts in one room, and every child gets both. That’s a bonus!” However, the one teach one assist model is the least effective of all seven models.

Honigsfeld and Dove describe seven co-teaching models. I am not going to unpack all of them here, but I want you to see the range, because the range is the point.

Some models keep the whole class together: one teacher leads while the other teaches on purpose (not assists, teaches); both teachers teach the same content side by side; one teaches while the other assesses. Some models split the class into two groups: both teach the same content in parallel; one pre-teaches while the other teaches; one re-teaches while the other extends. Lastly, one model puts students in multiple groups with both teachers monitoring and teaching.

While there are seven models, most co-teaching pairs I meet use one - one teach and one assist. They didn’t choose it because of its success rate. They only chose it because there wasn’t a choice.

When teachers become fluent in these seven ways, co-teachers meet and ask, “What is the purpose of this lesson, and which model serves that purpose?” The effective teams rotate who leads, so that nobody drifts into the aide role. That takes practice, and practice is something teachers do together, not something they just read about.

When I built JUNTOS, I wanted a framework a co-teaching pair could hold in their heads on a Tuesday at 9:40. So I built it into the word itself. Juntos means together. Each letter is a commitment.

J: Join as equals. Parity from day one. Both names on the board. “We” and “our” language when you talk to students. Both teachers are teaching all students, not “your kids” or “my kids” only “our kids.”

U: Understand our learners. The ML teacher sees things the lead teacher cannot, and the reverse is just as true. The team can use this to create asset-based profiles of their multilingual learners together. The language lens and the curriculum lens help create a clearer picture of each individual learner.

N: Negotiate roles. Choose the co-teaching model to fit the lesson’s purpose. Don’t allow habit to dictate. When you rotate who leads, you develop a teaching community. This may be the most challenging (e.g., discussing boundaries with other teachers). This is also the one that leads to the most growth.

T: Teach together. Every lesson carries a content objective and a language objective, and both teachers teach both. This is where co-teaching and sheltered instruction become the same conversation. The lead teacher owns the language objective too. The ML teacher owns the content objective too.

O: Observe and assess together. The shared process of designing assessments, sharing analyses, and grading decisions helps teaching become less accidental and more intentional. If only one is looking at the data, only one is teaching. Both teachers have a unique lens through which they see the class.

S: Sustain the partnership. There should be some kind of protected weekly co-planning. At the very least, these are five-minute reflection huddles at the end of each day. A partnership without planning time is a partnership in name only. Seeking regular times to plan as a team boosts teaching effectiveness.

None of these commitments is complicated. With the right training and the right tools, they are attainable for all teachers.

Taken together, the six commitments turn a schedule into a partnership. The push-in teacher stops being a visitor and becomes a co-owner of the classroom. The lead teacher gains a colleague who shares the planning, the teaching, and the results. Most importantly, the multilingual learners get two experts who are actually teaching them. That is what push-in promised in the first place.

Here is what I know after 30 years and 50,000 educators: you cannot learn co-teaching from a slide deck. Sit-and-get PD about collaboration is a contradiction in terms.

That is why ELI JUNTOS is live and experiential. Participants experience the seven models from the student seat. They practice the conversations that negotiating roles requires. They examine the tools that help push-in teachers and their classroom colleagues plan and teach together. Every activity in the course is itself a research-backed teaching move, named as it happens, so participants leave with co-teaching skills and a refreshed toolkit for their own classrooms.

For the co-teaching pair. ELI JUNTOS: Co-teaching for Multilingual Learners is the course built around the six commitments above. It is designed for push-in ML, ESL, and EL specialists and the classroom teachers who share their rooms.

For the partner teacher. The T in JUNTOS asks both teachers to teach both objectives. For most ML specialists, that is second nature. For most lead teachers, a language objective is new territory. ELI I and II give a classroom teacher the SIOP framework and 20 or more strategies to hold up their half of the lesson. A co-teaching pair where both teachers have been through ELI is a pair that speaks the same language.

For making it last. One live day builds the partnership. A structure afterward keeps it from drifting back to one teach, one assist by November. iCOACH peer coaching gives a co-teaching pair that structure, and this post on why 90 percent of professional development never reaches the classroom explains why it matters.

For the research. Andrea Honigsfeld and Maria Dove’s work on collaboration and co-teaching for English learners is the foundation for the seven models described here. Their books are worth every co-teaching pair’s time.

For your district. JUNTOS is also available as an on-site workshop. Contact us to talk about bringing it to your ML team.

Asi se puede mas. Together, more is possible.

ELI JUNTOS: Co-teaching for Multilingual Learners runs live on Zoom. Upcoming sessions:

•          October 15, 2026

•          December 8, 2026

•          January 25, 2027

•          March 5, 2027

Individual educators register for $175. Districts can register their ML team together for $1,500 for up to ten educators. Learn more and register.