Roughly a quarter of a second of video latency is enough to swallow a correction. That's the window an online language teacher has to catch a mispronounced vowel, feed back the right shape, and let the student try it again before the moment is gone. Most platforms burn through it before the audio even reaches the other end.
That tiny delay is where the feedback loop starts to fray. And when the loop frays, students stop improving. Not on day one, when everything still feels new, but around month three, when the honeymoon ends and the plateau shows up on schedule.
The First Two Months Flatter Everyone
Picture a student eight weeks into weekly online lessons. She can order coffee, introduce herself, describe her weekend in the past tense.
Her teacher is thrilled. She's thrilled. The reviews she leaves are glowing.
Then week ten rolls around and something shifts. The same errors keep coming back. New vocabulary slides off.
She books fewer sessions. By month four she's telling friends the app or the tutor "stopped working" for her.
Nothing stopped working. She hit the intermediate wall that language researchers have documented for decades, the point at which the fast early gains slow to a crawl and progress depends on a very different kind of practice than what got her there.
The Plateau Is a Feedback Problem, Not a Motivation Problem
It's tempting to blame the student's drive. Easier still to blame the schedule, the app, the second job. The real culprit is usually the loop between attempt, correction, and next attempt, the loop that carried her through beginner grammar and then went missing somewhere along the way.
Ericsson's work on deliberate practice puts it bluntly: expertise depends on a tight, self-reflective feedback cycle, and when that cycle breaks down, learners hit what researchers call arrested development. Language learners are no exception. Early on, every sentence produces obvious feedback, since the barista either hands over coffee or looks confused. Later, the feedback goes subtle, and video lessons often fail to deliver it.
For teachers rebuilding that loop, Harcourt Health put together advice for online language teachers that pairs well with any tooling decision. The tech only helps if the pedagogy underneath it is designed to keep students producing language, not consuming it.
Video Strips the Correction Out of the Lesson
Go back to that student. Her tutor hears a wrong preposition, waits for her to finish the sentence, then offers a gentle recast three exchanges later. On video, with a half-second round trip and a shared screen in the way, that recast lands on a sentence she's already forgotten.
Research on corrective feedback in second language acquisition is clear that timing and modality matter. Recasts, metalinguistic prompts, and explicit correction each do different work, and the medium either supports them or dilutes them. Video dilutes them by default. The teacher who was surgical in a classroom becomes vague on Zoom, not from lack of skill, but because the tooling absorbs the precision.
Tooling Has to Serve the Loop, Not Sit on Top of It
By month three, the student in this example isn't drowning in a lack of content. She has flashcards, a workbook PDF, a shared Google Doc, a chat window, and a tutor juggling all of it while trying to listen. Every tool competes for the tutor's attention, and the correction loop loses.
After month three, what moves her forward is a smaller set of well-chosen tools that keep the loop tight, not more content or a slicker interface:
- One capture surface. A single place where the tutor writes the correction the moment it happens, visible to the student in real time, instead of a chat scroll they'll lose.
- Asynchronous review. A short recording of each correction the student can replay before the next session, so the loop doesn't reset every week.
- Prompt banks, not slide decks. Materials the tutor can pull mid-conversation without breaking eye contact or sharing a screen that hides the student's face.
- Latency-aware pacing. A tutor who waits an extra beat after each prompt, because the platform has already eaten the first one.
What Month Four Looks Like When the Loop Holds
The same student, on a redesigned track, doesn't suddenly leap into fluency. She grinds. She hits the plateau, everyone does, but she notices herself hitting it, and the tutor notices with her.
Corrections land inside the sentence that produced them. She replays them between sessions.
Progress after month three is usually slower than progress before it. The students who keep improving are the ones whose feedback loop survived contact with video, not the ones with more talent or more time.