← Back to XLEMETY
Metrics
Every session on XLEMETY is led by a real person — a human Coach Teacher or a real native speaker. Here's what's human and what's automated in how you're measured.
Who you actually talk to: In Coach Mode, you're speaking live with a real, non-native Coach Teacher — never a bot, never a scripted character, never an AI voice. Your Coach Teacher corrects your pronunciation, grammar, and word choice live, in real time, right in the middle of the conversation — the same way any real teacher does. That live correction is the core of what you're paying for, not an afterthought. In Real Mode, Team Challenge, and XLEMETY Live, you get the same live correction from a real native speaker. The only two parts of XLEMETY that are fully automated, with no human on the other side, are the Placement Test and the Level Exam — both used purely to measure your starting point and your progression, not to teach or evaluate you as a person.
Being upfront about the numbers below: Beyond what your Coach Teacher or native corrects live in the moment, XLEMETY also tracks objective metrics — grammar patterns, pacing, vocabulary range — computed by automated analysis of your real, human-led session. These exist to give you a consistent way to see your progress trend over time, session after session. They're a second layer on top of your coach's live correction, not a replacement for it, and not a certified linguistic assessment.
Coach Mode Metrics
Fluency Score
How smoothly you spoke — pacing and minimal unnecessary pauses, based on your speech transcription's timing. It reflects rhythm, not vocabulary or grammar.
Grammar Score
Your Coach Teacher corrects grammar mistakes live, as you speak, in every session. This score is a separate tracking layer — checked against a rule-based grammar engine built specifically around common patterns Spanish speakers run into — that shows you the trend across sessions.
Pronunciation Score
Your Coach Teacher corrects your pronunciation live, in real time, as you're speaking — that's the actual correction that improves you. This score is a secondary trend indicator estimated through phonetic analysis of your transcribed speech, so you can see the trajectory session over session.
Intent Score
Whether your response actually addressed what the scenario was asking — communicating on-target, separate from whether the grammar was perfect.
Vocabulary Score
Range and appropriateness of the words you used relative to what the scenario called for.
Slang Score
Whether you used the informal, everyday expressions that specific scenario was designed to practice.
Phrasal Score
Same idea, focused specifically on phrasal verbs (e.g. "figure out," "run into") that scenario targeted.
Retry Rate
How many times you needed to retry a step (capped at 2 per step). This is a data point about difficulty for you personally — not automatically a bad sign.
Correction Success
Whether you incorporated a correction given earlier in the same conversation — a signal of real-time learning, not just first-try performance.
Conversation Flow Score
Whether your responses connected naturally to what was just said — real back-and-forth, not isolated correct sentences graded in a vacuum.
Confidence Score
Estimated from speech pacing and hesitation patterns in your transcription. It's an approximation of composure, not a psychological measurement.
Live Sessions (Real Mode, Team Challenge, XLEMETY Live)
Native's live feedback
The qualitative evaluation of how you actually did comes directly from the real native speaker at the end of the session — not from an algorithm. The same objective metrics above (grammar, vocabulary, pronunciation, flow) run quietly in the background as supporting data, but never replace what the native tells you.