89%
hit proficiency target
vs. 44% in class
Jochum, 2014
7 wk
= 3 semesters
intensive speaking immersion
Middlebury
0.70
effect size (d)
across 52 studies
Bryfonski & McKay, 2019
Output
beats input
production drives fluency
Swain, 1985

In a normal 50-minute lesson an individual student speaks the target language for only about 30 seconds; Gakimas gives every student continuous talk time across the whole session.

The talk-time gap

A student speaks for about 30 seconds in a 50-minute class.

Per-student target-language output, one lesson

Normal class ~30 sec

Teacher talk dominates; the class shares ~24% of talk time between everyone.

Gakimas session the whole session

Every student speaks continuously, on the teacher's own curriculum, every day.

100× more time on the student's own voice.

Source: DESI study (DIPF, 219 schools / 11,000 students) found students together spoke 23.5% of class time; per-student estimate via Perez Castillejo, 2017.

The testing effect

Retrieval beats review.

Re-reading and re-listening feel like learning. They don’t last. Pulling the language back out of your own head — producing it out loud — is what makes it stick. Every spoken turn in Gakimas is an act of retrieval.

Percent of material recalled on a final test, one week after learning — Karpicke & Roediger, Science (2008)
Retrieval vs. re-study: one-week retention After the same initial learning, students who kept retrieving recalled about 80 percent a week later, while students who kept re-studying recalled about 36 percent. ~80% Kept retrieving ~36% Kept re-studying
0% forgotten after a week by re-studiers — vs. just 13% by those who kept retrieving Roediger & Karpicke, 2006
0% of 159 studies favored retrieval over re-study (mean effect g = 0.50) Rowland meta-analysis, 2014

Listening to a tutor is review. Speaking back is retrieval. Gakimas makes every student do the part that lasts — out loud, every day.

The affective filter

Lower the stakes.
Raise the output.

Speaking anxiety is the single biggest brake on classroom production. Around one in three language learners carries it — and speaking is the skill it hits hardest. Take the judgment out of the room and the brake comes off.

1 in 3
language learners experience foreign-language anxiety — and speaking is the most affected skill.
A non-judgmental partner lowers the affective filter — raising willingness to communicate.
Mechanism, illustrative: remove the audience and fear of evaluation, and more students choose to speak.

Mechanism, not a Gakimas-measured outcome. Anxiety — chiefly communication apprehension and fear of negative evaluation — suppresses second-language output; lowering it raises willingness to communicate. Horwitz (2001); Horwitz, Horwitz & Cope (1986); MacIntyre, Clément, Dörnyei & Noels (1998).

The mechanism

Corrected in the
moment.

When a tutor reformulates a slip in the same breath the student spoke it, the correct form lands while the sentence is still warm. Linguists call it a recast. Gakimas runs one on every turn.

  1. Student

    Yesterday I go to the market.

  2. Tutor

    Nice — yesterday you went to the market. What did you buy?

    recast · same turn
  3. Student

    I went for fruit and bread.

    uptake · corrected form reused
A slip, a gentle reformulation, the right form reused — all without breaking the conversation.

Corrective-feedback effect on L2 learning

d = 0.64

A durable medium effect across 33 studies — and it holds over time.

every turn

A teacher fielding 25 voices can recast a fraction of slips. Gakimas gives each student a recast on every spoken turn.

Effect size: Li (2010), Language Learning 60(2), 309–365 — meta-analysis of 33 studies. Recast & uptake defined in Lyster & Ranta (1997), SSLA 19(1). Li 2010 Lyster & Ranta 1997

More hours. Lower ceiling.

Roughly the same hours, worlds apart. Four years of classroom seat time plateaus around B1; the same time spent speaking reaches C1+. The ceiling isn’t the hours — it’s how many are spent talking.

B1
~600 hrs
of classroom time4 years · ACTFL
C1+
~600 hrs
of speakingFSI intensive

Four years of U.S. high-school study lands most learners near B1 (ACTFL proficiency benchmarks). The Foreign Service Institute reaches professional working proficiency (≈ C1) in ~600 intensive hours — comparable time, far higher ceiling, because it is spent speaking.

Real practice. Day one.

0
students speaking
0
conversation turns
0min
of real speech
0
words spoken
0wpm
words per minute

Two demo sessions, Mercersburg Academy — May 2026.

See it on your curriculum.

Book a pilot