Summary

word and identifier are defined with the same regex:

identifier: () => /[a-zA-Z_][a-zA-Z0-9_]*/,
...
word: () => /[a-zA-Z_][a-zA-Z0-9_]*/,

This is a deliberate-looking ambiguity, and it is listed in conflicts:

conflicts: ($) => [
  [$.source_file, $.typed_variable],
  [$.identifier, $.word],
],

The two are used in different contexts (identifier for $ref variable references, word for names), but since they match identically, the conflict forces tree-sitter to try both productions. This generates spurious ambiguity and can produce surprising parse trees.

Expected

Pick one canonical token and delete the duplicate. If identifier and word must remain semantically distinct (e.g. for highlighting or transpiler consumers), give them different productions or mark one as an alias.

Files

• grammar.js — identifier, word, conflicts