The Challenge of Analyzing a Dynamic Text

[rikforto]

but a different finding in Pinyin

If "words" are the pinyin romanizations of Chinese, 100% of all word types form a single connected network through single-letter edits.

You are not allowed to view links. Register or Login to view.

This highlights another distinction; Chinese words in hanzi have an edit distance one independent of text. I might be wrong here, and frankly it's the abstraction from this point that I care about, but I think it should be possible to construct a text of some length that has word with minimum edit distance 2 apart. For initials, you'd need a lacuna in the alveolars and the finals are prolific enough it should take a hot second for the text to converge on edit distance 1. You are absolutely correct for the lexicon. Bopomofo might converge must faster, on reconsideration.

Either way, bigger point: This is dependent on the language, the representation, and the sample in ways that mean you can't simply infer the language or lack thereof.