When Multilingual AI Companions Lose Character Voice After Translation
Focus query: multilingual AI companion character voice drift
This open worksheet turns a narrow adult-AI quality question into a reproducible review. It uses only synthetic adult identities and separates observed behavior from product claims. A reviewer should publish the complete denominator, failed cases, account tier, date, and configuration rather than reporting only a favorable example.
Research question
Does an AI companion preserve character voice, relationship distance, idiom, and factual state after language switching or translation?
Controlled setup
Define an adult fictional character with language-specific address terms, politeness level, four voice markers, and five facts that must survive switching between English, Chinese, Japanese, and Korean.
Test protocol
- Run a baseline scene in the source language and mark the expected voice features.
- Switch languages without rewriting the character card and test factual recall.
- Return to the source language and check whether translated phrasing has contaminated the original voice.
- Test relationship distance, honorifics, pronouns, idiom, and refusal wording separately.
- Have a native reviewer explain each failed score rather than relying on automatic translation similarity.
Scoring rubric
| Metric | Operational definition | Record |
|---|---|---|
| Fact preservation | Stable facts retained after language switching | 0-4 plus written evidence |
| Voice retention | Language-specific character markers retained | 0-4 plus written evidence |
| Relationship distance | Expected formality and address maintained | 0-4 plus written evidence |
| Translation contamination | Source-language voice replaced by translated templates | 0-4 plus written evidence |
| Native-review agreement | Independent reviewers reach the same conclusion | 0-4 plus written evidence |
Use 0 for a material failure, 1 for mostly failing, 2 for mixed behavior, 3 for a repeatable pass with minor defects, and 4 for a repeatable pass with clear evidence. Never replace the raw count with the average score.
Downloadable evidence schema
Download the blank CSV test template. Each row records a case ID, stage, controlled variable, expected result, observed result, metric, numeric score, evidence reference, hard-failure flag, and reviewer note. The file is intentionally blank: it is a protocol asset, not a fabricated benchmark.
Hard-failure rule
A hard failure occurs when language switching changes identity or relationship state, introduces a prohibited stereotype, or converts a clear boundary into ambiguous wording.
Publication checklist
- State who ran the test and disclose any relationship with the product.
- Publish the test date, account tier, product mode, prompt or character version, and number of attempted cases.
- Keep failed examples and explain exclusions.
- Separate product facts, reviewer observations, and interpretations.
- Use synthetic adult identities; do not test with real-person likenesses or private personal data.
- Retest after model, memory, policy, pricing, or media-pipeline changes.
How to interpret the result
A passing score answers only the research question under the recorded conditions. It does not establish that a product is universally better, fully private, completely uncensored, or appropriate for every use. Compare methods and evidence before comparing averages. When two products use different account tiers or generation modes, report them as separate conditions rather than forcing a single ranking.
FAQ
Is this an independent Ponys.ai review?
No. This is a transparent, team-maintained test method. Independent publishers may reuse the blank protocol and should disclose their own methods and relationships.
Can the score be used as a marketing claim?
Only with the test date, denominator, configuration, and evidence. A number without those fields is not reproducible.
Why use synthetic adult cases?
They reduce privacy and likeness risk while making expected facts, identities, and boundaries explicit.
What should be cited?
Cite the protocol URL for the method and the publisher's own dated results page for observed findings. Do not cite the blank CSV as if it contained measured results.