SouCha is a Go package that allows a user to simulate historical sound changes. Provided a list of sound changes ("rules"), SouCha can apply each change sequentally to a set of words or phrases.

All logic is contained in package parser. Create a new language with l := parser.NewLanguage("__rules to parse__"). Then you can evolve a segment with l.Evolve("word/phrase").

Rules must be in the form initialSound > finalSound / preCondition _ postCondition. If conditions for the sound change don't exist, then you can omit everything after the slash. Any component can be omitted if it is the empty set ∅ (it is non-restrictive).

Any of these four components can contain:

• Spaces and commas, indicating a break between two parts of a set: x y and x,y both indicate the set of x or y.
  • To indicate a sequence of consecutive phonemes, put no spaces in between: xy indicates xy in sequence.
  • These consecutive phonemes can also be sets of phonemes: x{y z} indicates xy or xz.
• Curly braces {}, indicating any one of a set: {x y} allows you to group phonemes.
• Parentheses (), indicating an optional phoneme: (ʰ) indicates either ʰ or nothing.
• Square brackets [], indicating a combination of named sets, with their own rules inside:
  • +setname and &setname indicate an intersection, meaning the matched phoneme must be in that set.
  • -setname and !setname indicate the complement, meaning the matched phoneme cannot be in that set.
  • |setname indicates a union, meaning the matched phoneme can be in this set or the set before it.
  • () indicates logical groupings, NOT optionality– use |setname for optionality instead.
  • {} indicates an anonymous set; it acts as a named set outside of the braces, but the symbols inside are assumed to be phonemes, not named sets.
  • NAME: at the beginning of square brackets indicates the set's name. It can be any token with no spaces and referenced later using [NAME:] like any other phoneme to describe the specific phoneme matched previously.

# is a special character in preCondition and postCondition, indicating a word boundary (beginning/end of a word).

A named set must be in the form setname = x y z, either space or comma-seperated. They cannot contain other named sets, curly braces, or parentheses as of now. They represent a set of phonemes.

There are many sound change appliers online already, but SouCha allows for a rule input that is closer to general linguistic conventions. Rules might look like

[V:+vowel] > / [V:] _ {x y}

I.e, vowels disappear when the same vowel appears before and is followed by /x/ or /y/.