Domain review: C1 minor_issues as the explicitly thresholded CLICS4 result; C2 minor_issues as the descriptive two-hit result and nonsignificant class test. Significance minor for each: useful lexical inventory, but not an exhaustive test of embodied cognition or psychological mechanics. I read the paper, plan, code, inputs specification and results, independently reproduced the complete R1 in the earlier assigned reproduction, and audited its pair coverage.

C1's finite operational result is supported: no eligible non-load pair passes BOTH percentile>=.95 and>=3 colexifying families. Its broader title, Summary and Discussion are not supported. No qualifying hit is not no colexification: examples include PUSH-WANT and PULL-WANT in2families, RUB-THINK in2, and several1-family pairs beyond the95th percentile. The manuscript acknowledges some but still says senses are kept apart as words and weight is the only mechanical domain lexicalized psychologically. Those sentences erase the threshold distinction. State no pair passed this prespecified screen, and restrict conclusions to these concepts and this dataset.

Coverage audit:42 total anchors including5load anchors times30counted targets gives1260potential pairs; R1 actually evaluates1102. SAD has eligible comparisons for only11of42anchors. ANXIETY,REGRET,GUILTY,DOUBT and DIFFICULT each have39. Missing/ineligible pairs are not observed non-colexifications and should be visible in a coverage table with available families per pair and each class's tested denominator. C1 should say eligible pairs with at least30jointly covered families. The class null means range.353(path) to.719(flow), not.48-.72 across all four classes; correct that numerical summary.

CLICS records elicited concept labels and forms, not exhaustive dictionaries or word senses. A shared normalized form can represent polysemy or homophony, and absent shared forms can reflect limited elicitation, translation choices, tokenization or inflection rather than semantic separation. Thirty-seven concept glosses are not37universal lexical items. Grouping push/pull/swell/break as normal stress and fast/slow/boil as flow/momentum is a hypothesis about semantic representation, not a validated physical-psychological measurement scale. The database lacks the key nominal pressure/stress/tension concepts as the paper acknowledges. Linguistic mechanisms involving phrases, derivation or grammar are outside this test; their exclusion prevents an inference that deep embodied mappings should have appeared as lexical identity.

The family statistic gives a hit to a family if any sampled variety shares a form. Families with more varieties/forms have more opportunities to hit. It reduces direct counting of related languages but does not make families independent experimental replications or equalize detection opportunity. Unknown-family varieties are separate isolate IDs in code, potentially treating multiple resources as separate independent families. Areal contact, shared wordlist sources and duplicate varieties deserve checks; macroarea listing alone is descriptive. Random anchors are restricted to>=60family coverage, while target-pair eligibility is>=30; matching class size alone does not match coverage, word class, polysemy or elicitation opportunities. Use coverage/degree/part-of-speech matched nulls or label the existing null as an operational comparison.

Upper-tail class tests ask about enrichment. With0hits they give p1 by construction; they do not test a deficit, equivalence or absence of meaningful excess. With expected counts below1, absence of hits is unsurprising. Neither the known-positive benchmark's p.0633 nor its nonsignificance quantitatively measures power. Retain the preregistered warning, but estimate sensitivity under explicit alternatives before calling the test insensitive as an established finding. Reporting percentile>=.95 is a rank criterion, not a pair-level significance test with familywise error controlled across42x30pairs.

C2's2hits are supported as forms in the dataset: HEAVY-DIFFICULT7/57families; LIGHT(WEIGHT)-EASY11/50; the latter across5macroareas. These are effort/ease concepts, not direct evidence about emotion or depression. The anchors were known positives before registration, honestly disclosed, so this is a benchmark rather than a novel discovery. Classp.0633 does not establish significant superiority to other mechanical classes; that contrast was not tested. Preserve the numerical results and uncertainty without universalizing weight as the sole domain.

Preregistration and transparent data/code make the operational screen auditable. Revising the scope, coverage accounting and null interpretation would retain that contribution. No source identity was sought or learned, and no claims about human psychological mechanisms are warranted by this lexical screen alone.