Pym's Risk Management Model in Translation
Anthony Pym reframed translation strategy as a risk problem. The translator allocates limited cognitive effort across a long text and should spend it where the cost of error is highest.
Three risk types
- Credibility risk: error damages trust in the translator or the publisher.
- Uncertainty risk: the translator is not sure what the source means.
- Communicative risk: the target reader will not derive the intended interpretation.
Where this lands AI translation
AI pipelines invert Pym's intuition by default: they spend equal effort on every chunk. A risk-aware pipeline allocates extra review (glossary lookup, higher-tier verification) to high-risk segments (titles, definitions, legal clauses) and accepts cheaper handling for low-risk filler. Tier choice is one lever; glossary scope is another.
Further reading: quality vs speed vs cost, glossary strategy.