inst/NEWS.md

In ConversationAlign: Process Text and Compute Linguistic Alignment in Conversation Transcripts

ConversationAlign 0.3.1

- 2025-7-29

Breaking changes

This is a minor release involving one bug repair

Modified

• prep_dyads() added several missing cleaning functions that had been inadvertantly omitted during an earlier merge in preparation for CRAN

ConversationAlign 0.3.0

- 2025-7-21

Breaking changes

This is a minor release involving mostly optimization of functions.

Added

• Software license now LGPL 3.0

Modified

• Omitted prior GPL software license
• Path to Journal of Open Source Software (JOSS) now points to main

ConversationAlign 0.2.0

- 2025-7-15

Breaking changes

This is a major release. We have included many new features and fixed numerous bugs.

Added

• Added news (you're reading it now), discussions, and issues/bug tracker to Github repo
• prep_dyads() that combines two previous preparation steps clean_dyads() and align_dyads()
• New function read_1file() that formats a conversation transcript already in your R environment
• Improved handling of contractions with error detection/substition of non-ASCII character apostrophes. Contraction expansion (e.g., 'they're' to 'they are') now handled by a separate internal function call using custom regex.
• Added argument omit_stops to prep_dyads specifying optional stopword removal
• Added argument which_stoplist to `prep_dyads`` specifying a four stopword list options (SMART_stops, MIT_stops, Temple_stops25, CA_orig_stops).
• Added pdf manual for ConversationAlign
• Validation checks to read_dyads() will throw error and warning message if any transcript has more than or less than two participants for any individual conversation (Event_ID)
• Added corpus_analytics() function that produces a variety of descriptive summary statistics (e.g., n-conversations, type token ratio, average turn length) in a table format.

Modified

• summarize_dyads() restructured in significant ways including the addition of three new arguments: custom_lags default lags for correlation are set at -2.0,2 users are free to specifiy additional lags , default is NULL, sumdat_only produces a summary dataframe with values averaged to two rows per conversation (one for each participant, corr_type specifies correlation to apply to lagged data
• Resampling of AUC in summarize_dyads() has been omitted in favor of proportionally rescaling dAUC to a standardized/fixed number of turns (100)
• Modified internal lookup datase lookup_db, replaced with lookup_Jul25, includes many more dimensions and options for scaled vs. raw versions of variables
• Moved all internal data to a different repository (ConversationAlign_Data) in order to get the package within CRAN file size constraints. Package now loads data from external HTTPs source
• Omitted outside calls to libraries from functions

Fixed

• Warnings and checks (variable names, etc) to read, clean, align functions

Deprecated

• We are in the process of phasing out our original stopword list from earlier versions of ConversationAlign (pre 2025). We have included the old list as an optional call to the argument clean_dyads() as CA_OriginalStops. However, the default stopword list is new - Temple_Stops25
• Eliminated several dimensions from lookup database: prevalence, hope, stress, politeness, empathy, prevalence, closeness, encouragement, hope, doubt, hostility, surprise
• clean_dyads() and align_dyads()

ConversationAlign documentation built on Aug. 8, 2025, 7:22 p.m.