Corpus Analysis: Why?

Hi! Something that has always puzzled me is how type designers use corpus analysis in their process. How does knowing the frequency of glyphs, or glyph pairs influence your decision making?
In the non-Latin script arena, I can see that information as being useful if you're trying to determine which conjunct glyphs (via GSUB) to include in your font. But what else do you find it useful for? Just curious :)

Comments

  • I imagine it'd be useful for strategizing ligatures in a connecting script.
  • I’ve mostly used it regarding conjuncts in Devangari. And occasionally I use it to try and find out if and how a character is used when documentation of a character in a certain language is lacking.
  • I rarely look at pair frequency anymore, and then only if I am wondering if I need bothering to kern a rarely if ever used combination.
  • I use pair frequency in the case of textured designs where, instead of using cycling alternates, I use ligatures. For example, designs with scratches, splatters or pattern elements that span 2 or 3 letters. I used pair & triplet frequency charts from several languages to compile a master list.
  • It can help deciding what kerning exeptions one should add. 
  • I used 'em to design early phase sketch words; vowels + most common consonants