Changelog • joeyr

joeyr 0.11 (July 10, 2025)

Added fave_to_wells. Not as rubust as switch_transcriptions and its shortcut functions like arpa_to_wells, but I wanted to get something in since I find myself copying that code over and over now anytime I analyze FAVE data.

The normalization functions (norm_logmeans, norm_anae, and norm_deltaF) have been depreciated for several reasons. The tidynorm package handles these much more effectively and efficiently than what I was able to do (see introduction here). The three functions now do nothing except give a message to the user recommending that they switch to tidynorm instead. For those who don’t want to make the switch, the three functions will still exist in this package but have been renamed joeyr_norm_logmeans, joeyr_norm_anae and joeyr_norm_deltaF. This prevents conflicts as well.
I’ve added a very simple OoO function that reminds me what my recommended order of operations is.
I’ve also removed the This is the "joeyr" package message when loading.

Updated code_allophones
- MOUTH gets a prelateral allophone now: PROWL. This means the elsewhere allophone is now called BOUT.
- PRICE gets a prelateral and a prerhotic allophone: CHILD and PRIOR.
- Prevoiceless PRICE has been changed from PRICE to BITE. Prevoiced PRICE has been changed from PRIZE to BIDE.

Fixed two issues with GOOSE allophones in code_allophones.
- Words like tool which are both post-coronal and prelateral, were previously coded as post-coronal. That has been fixed so that they’re not prelateral.
- There wasn’t a prerhotic allophone of GOOSE, which caused words like demure to be erroneously classified as TOOT. Now, prerhotic GOOSE is CURE. Note that prerhotic FOOT is also CURE.

norm_logmeans() has a raised cap before triggering a warning message.
find_outliers() now as the verbose argument which, by default, suppresses the warning message triggered when there is not enough data.

code_allophones() now added.
norm_logmeans() now added, which does a log-means normalization based on Barreda & Nearey (2018).
norm_anae() updated. The use_telser_g argument has been changed to simply g. Previously, if you wnated to calculate g internally, you’d do use_telsur_g = FALSE and that code didn’t work. Now, you’d say g = "calculate", and that does work this time. Additionally, some new functionality is that g can be set to any arbitrary value, such as zero: g = 0.

arpa_to_keywords() has been replaced with the more robust and flexible switch_transcriptions.

norm_anae() now uses the Telsur G by default. This can be suppressed and a new G value can be calculated using the data (the old default behavior), by specifying use_telsur_g = FALSE.
norm_anae() now requires the data to be grouped by speaker prior to running. This makes it more consistent with the other normalization procedures in this package.
wells_to_arpa() was created, as an inverse function to arpa_to_wells() because I needed it one time.
norm_deltaF() has been fixed so that including F4 now works. Additional arguments have been added as well.

pillai() and tidy_mahalanobis() get much clearer examples and much more documentation.
arpa_to_keywords() no longer crashes if a dataset didn’t have all vowels present.

Switching to pkgdown, which involved some overdue tidying of the package
- Removed depreciated functions, including hz, bark, spread_n, move_x_before_y, move_x_after_y, and all the functions related to the old outlier detection procedure (joey_filter, joey_do_pca, joey_adjust_cutoff, joey_rm_outliers).
- Removed the datasets. They were just added like a week ago, but I decided to move them to a separate package joeysvowels.
- Renamed the .R filenames
Added a NEWS.md file to track changes to the package.