Nu toate indiciile cântăresc la fel

If a deep relationship is real, surviving cognates should be disproportionately concentrated in very stable meanings rather than randomly distributed across the lexicon.

Evidence: The proposed 'patterning' method uses differential lexical replacement rates (after Dyen) as a diagnostic.

Inference: Non-random clustering in low-replacement meanings is treated as evidence against chance.

methodological_proposal · methodological

A useful basic-word list for reconstruction should be optimized empirically for retention across language families rather than inherited mechanically from a canonical Swadesh list.

Evidence: The author describes a '100 best kept words' list derived from cross-family retention and proposes starting family reconstruction with it.

Inference: Lexical stability is treated as an empirical property of meanings.

author_proposal · methodological

Substrate vocabulary is not semantically random: across languages it clusters in recurrent meaning domains.

Evidence: French and Romanian substrate sets are compared side by side and yield clustering in similar semantic fields.

Inference: Some meanings are more likely than others to survive population/language replacement.

plausible_but_requires_independent_replication · empirical

Some meanings are geographically stable even when the language changes, which helps explain why particular lexical slots disproportionately preserve substrate words.

Evidence: The conclusion explicitly contrasts meanings associated with substrate and meanings easily replaced.

Inference: Semantic function can influence lexical survivability across language shift.

plausible_but_requires_independent_replication · interpretive

The strongest surviving idea in the 2010 paper is its a priori stability-patterning principle: sparse deep similarities are more informative if they concentrate in independently stable meanings, but this must be paired with an explicit chance model.

Evidence: Pagel et al. 2013 later argued for ultraconserved high-frequency words; Mahowald & Gibson warned that short frequent words can also appear related by chance.

Inference: Stability and false-positive calibration are complementary, not competing, requirements.

editorial_assessment · methodological

Modern typology shows that nasal/basic consonants are common in personal pronouns and that some pronominal consonant patterns have broad areal distributions not explained by deep genetic relationship.

Evidence: Nichols & Peterson describe m-T and n-m systems as geographical spreads and explicitly reject deep genetic relatedness as their general explanation.

Inference: A generic consonant-frequency null is inadequate for pronoun consonantism.

editorial_assessment · methodological