language: en
script: Latin

normalization:
  unicode_normal_form: NFC
  lowercase_for_indexing: true
  # Surface forms preserve original casing.
  # Indexing and lemma lookup use lowercased forms.
  strip_diacritics: false
  # English retains diacritics in loanwords (naïve, résumé).

punctuation:
  strategy: preserve
  # Punctuation is preserved in normalization output.
  # Lift rules are responsible for ignoring or consuming it appropriately.

spacing:
  strategy: whitespace-split
  # Standard whitespace tokenization. No special boundary rules.

script_direction: ltr