cat-linguistic-style-matching

Compute function-word frequency profiles for the user's posts and for a community baseline excluding that user, then apply the LSM formula per category: 1 minus the absolute frequency difference divided by the sum. Average the nine category scores for a composite LSM.

LSM scores range from roughly 0.50 to 1.00. Scores at or above 0.80 indicate strong accommodation, 0.65 to 0.79 indicate moderate accommodation, and below 0.65 indicates weak or no stylistic convergence with the community.

Each text sample needs at least 50 words, the user needs 5+ posts and 200+ total words per community, and the community baseline needs 20+ posts from at least 5 unique authors excluding the target user.

Including the user's own posts in the baseline inflates LSM by measuring self-similarity rather than accommodation. The baseline must reflect only other members' writing to represent a genuine community norm.

No, not with the provided function-word lists, which are validated only for English. Non-English text requires validated function-word categories for that language, and machine-translated text should be avoided because translation alters function-word distributions.

LSM measures stylistic similarity, not social motivation, loyalty, or intent, and cannot distinguish conscious from unconscious accommodation. It also ignores lexical choice, syntax, tone, and discourse structure, and small baselines produce unstable scores.