Editing Openai/69186d20-2d14-8013-b9c6-c09000f173f6 (section)

=== Assistant: 2511141659ET ===
251114_1659ET

got it—that helps calibrate.

So in the “phylogenetic” sense you care about, I’ll keep distinguishing three cases:
# Shared innovation (true close kinship of the feature) - Same specific change in a common ancestor → inherited by both daughters. - E.g. if both languages show the same weird sound change that clearly happened once pre-split.
# Parallel exploitation of a shared background quirk - What’s happening with Old Irish fortis sonorants vs Welsh voiceless sonorants: - Both branches inherit from Proto-Celtic: - strong use of geminate sonorants; - tolerance for sonorant fortition carrying morpho-syntactic load. - Goidelic: geminates → fortis vs lenis (voiced, tense vs lax), then mostly collapse. - Brythonic (Welsh): some fortis sonorants → devoiced [ɬ, r̥, etc.], which then phonologise. - So they’re related in the sense: “same ancestral tendency → different branch-specific innovations.” Not a shared derived feature, but a shared substrate that each branch pushes in its own direction.
# Pure convergence / coincidence - Similar-looking outcome, but no special shared prehistory beyond “they’re both human languages with broadly similar articulatory options.”

You’re basically always asking:

: 

I’ll keep answering in that register: when two things “look similar,” I’ll explicitly say which of the three buckets they’re in and how far up the tree the common cause really is.