Translation as Probability Cloud - Against Premature Collapse in Cross-Linguistic Semantic Recovery

Scholarly translation in contested-term contexts collapses what should be a distributional commitment into a single point estimate, and the collapse becomes load-bearing across generations of secondary scholarship that cite the gloss as if it were the source term. This paper diagnoses the failure as a domain-specific instance of Binary Bias operating at the lexical level, proposes a probability-cloud apparatus that preserves the distributional structure of source-language pointers across translation operations, demonstrates the apparatus across four cases from different language families (Hebrew hesed, Greek logos, Chinese dao, Japanese ma) with brief outward reference to Old Norse hugr, and specifies what a retranslation paper looks like under the framework: a candidate set, conditioning variables, weight assignments with evidentiary trail, and perturbation-readiness as load-bearing methodological commitment. The framework is QEDA applied to philology and the methodological complement to Superpositional Text at the opposite end of the communicative pipe. It opens onto a research programme provisionally named in distributional form, with quantum philology, probability-cloud philology, uncollapsed philology, and hyperphilology as current candidates with conditioning-dependent weights. The framework asks no exemption from its own methodology for the methodology paper itself.