Why Accent Research Needs Self-Reported First Language
A model prediction is not a language label. Responsible speech data keeps prediction, self-report, page locale, and country separate.
Quick answer
Accent research needs a speaker’s self-reported first language because a model prediction is an output to evaluate, not ground truth. WitAccent asks before revealing the guess, supports selecting more than one language, and keeps country and interface language separate. A declaration describes the speaker’s history; it does not turn a prediction into fact.
Why should the prediction come second?
If the predicted class is copied into the answer field, later evaluation merely confirms the model’s own assumption. WitAccent asks for the declaration before revealing the guess and currently lets a person select one or multiple languages, keeping the human answer independent.
The answer is presented as saved only after the response is confirmed. Skipped or failed answers remain unlabelled rather than inheriting a prediction.
Why do exact labels matter?
Hindi and Urdu may share some speech features but are distinct declared languages. Cantonese and Mandarin have different phonologies and language codes. Filipino and Tagalog need clear wording rather than a vague regional class. Selecting multiple languages can describe an early bilingual history more honestly than forcing one choice.
A model may use broad classes for technical reasons. Those classes must not overwrite a speaker’s exact declaration.
Which fields should stay separate?
| Field | What it means |
|---|---|
| Page locale | Language of the interface |
| Country or region | Optional location context |
| Model prediction | Fallible acoustic output |
| Declared first language | Speaker-provided language history |
| Training eligibility | Consent and data-quality decision |
Frequently asked questions
Why ask before showing the prediction?
It reduces the chance that the model’s guess anchors the speaker’s answer.
Is country a substitute for first language?
No. Countries are multilingual, and languages cross national borders.
What if someone has more than one first language?
WitAccent supports selecting multiple languages so an early bilingual history does not have to be reduced to one label. That declaration remains separate from the model’s guess.
What happens when someone skips?
The sample remains unlabelled. A prediction must never be inserted as a self-declared answer.
