The comparative method reconstructs a proto-language by lining up cognate words across daughter languages and inferring the regular sound changes that connect them. More text, more known correspondence rules and more internally-consistent paradigms all push a reconstruction from speculative toward settled.
Reconstruction confidence:
words = corpus(k) * 1000 * 0.72 (transcription quality)
conf = words / (words + 5000) * (0.6 + 0.08 * 5 related langs)
Phonological coverage:
coverage = (phonemes / 42) * (1 - 0.18 residual ambiguity)
against 14 known sound-change rules
Grammar strength:
strength = (paradigms / 60) * (18 syntax patterns / 24) * 0.74 consistency index
- Corpus size — the effective transcribed word count feeding the comparative method; larger corpora at a fixed 72% transcription quality and 5 comparison languages push reconstruction confidence toward its ceiling.
- Reconstructed phonemes — how many of the proto-language's ~42 estimated phonemes have been recovered against the 14 known regular sound-change rules, discounted by 18% residual ambiguity that no corpus size fully resolves.
- Morphological paradigms — how many inflectional paradigms have been reconstructed, weighted by 18 attested syntax patterns and a 0.74 cross-language consistency index.
Each bar's height and glow track its metric in real time; taller, brighter columns mean the reconstruction is closer to what the comparative method can ever recover from fragmentary evidence — pushing any one slider to its extreme shows where the model saturates.