Changelog

Version 0.1.0

Released on: 2026/XX/XX

First public release of liwca.

• Added liwca.datasets.dictionaries.fetch_scope and liwca.datasets.dictionaries.fetch_psychnorms: per-stem fetchers that slice a single column from the SCOPE / psychNorms metabase into a weighted .dicx dictionary. Companion helpers list_scope_stems and list_psychnorms_stems enumerate the available stems.

• Breaking: liwca.datasets.tables.fetch_scope and liwca.datasets.tables.fetch_psychnorms now return the column-classification metadata tables instead of the full word-level score matrices. For the score matrices, use the new per-stem fetchers in liwca.datasets.dictionaries.

• Breaking: liwca.Liwc22 mode-method kwargs were renamed to be more pythonic and consistent with sklearn/pandas. The argument order in each mode method was also reorganised - identity (dictionary, text_columns, id_columns/id_column) comes first, behaviour knobs in the middle, output-shape near the end, and rare escape hatches (text, env_var) at the tail. No deprecation aliases - update call sites directly.

  Cross-mode column selectors:

  • column_indices -> text_columns (wc/freq/mem/context/arc).

  • row_id_indices -> id_columns (wc).

  • index_of_id_column -> id_column (mem/context/arc).

  Mode-specific renames:

  • console_text -> text (wc).

  • environment_variable -> env_var (wc).

  • category_to_contextualize -> category (context).

  • words_to_contextualize -> words (context).

  • output_data_points -> include_data_points (arc).

  • mem_output_type -> dtm_format (mem).

  • expanded_output -> expanded (lsm).

  • regex_removal -> remove_regex (ct).

  • prune_threshold_value -> prune_threshold (freq/mem).

  • omit_speakers_num_turns -> min_turns (ct/lsm).

  • omit_speakers_word_count -> min_words (ct/lsm).

  • segments_number -> n_segments (arc).

  • n_gram -> ngram (freq/mem).

  • speaker_list -> speakers (ct).

  • url_regexp -> url_regex (constructor).

  Behavioural simplifications (rename + type change):

  • word_window_left and word_window_right collapsed into a single word_window: int | tuple[int, int] (context). An int applies to both sides; a (left, right) tuple sets them independently (default: 3).

  • calculate_lsm: int (1/2/3) -> level: Literal["person", "group", "both"] (lsm; default "both").

  • output_type: int (1/2) -> pairwise: bool (lsm; False = one-to-many, True = pairwise; default False).

  • scaling_method: int (1/2) -> scaling: Literal["percent", "zscore"] (arc; default "percent").