In CoEDL/yinarlingi: Validation Functions for the Warlpiri Dictionary

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "README-"
)

yinarlingi

Overview

The purpose of yinarlingi is to provide a set of convenience functions for validating Warlpiri dictionary data. Essentially, it pre-configures various functions from the tidylex and tidyverse packages with parameters specific to the Warlpiri dictionary data, so that those within the Warlpiri dictionary project can quickly run and re-run frequently used routines (e.g. data validation, derived views), and also so that these tests can be run in a Continuous Testing/Continuous Deployment environment such as GitLab pipelines:

Yinarlingi is a Warlpiri word for echidna.

Installation

You can install yinarlingi from GitHub with:

# install.packages("devtools")
devtools::install_github("CoEDL/yinarlingi")

Examples

Test that a Warlpiri dictionary entry's data lines are well-ordered

The Warlpiri dictionary consists of a large plain text file of the form (where \me stands for main entry; see full list of codes):

\me jampaly(pa) (N) (PV): (La,Wi,Y)
\dm spatial: tactile: \edm
\gl sharp, pointed \egl
\rv sharp \erv
\eg
\we Karlangu ka karri jampalypa ngulaju yiri-nyayirni. \[@@]
\et The digging stick is sharp, that is very sharp pointed. \ewe
\eeg
\ant jampilypa, munju \eant
\cf jaarn-karri-mi, lirra jampalypa \ecf
\syn jalkarra, larrilpi, yiri \esyn
\eme
\sse jampaly-pi-nyi (V):
\def xERG cause y to come to be sharp (<jampalypa>) \edef
\gl sharpen, trim to ^point \egl
\rv sharpen \erv
\eg
\we Jampaly-pungulpalu wangkinypa, yangka kuja munju-jarrija. \[@@]
\et They sharpened the stone-axe which had became blunt. \ewe
\we Karlangulu jampalypa-pungu. \[Tiger Jakamarra]
\et They sharpened the points of the digging sticks. \ewe
\eeg
\csl YSL#1336, YSL#313 \ecsl
\syn yiri-ma-ni \esyn
\esse

...

A well-structured Warlpiri entry is defined as a Nearley grammar (see full grammar here), of which a snippet is shown below:

entryBlock       -> "org":? "dm":* "def":? "lat":? "gl":? "rv":? "cm":*
                      (exampleBlock:+ | paradigmExample:+):?
                      crossRefs

We can see here that a semantic domain line dm must come before the gloss line gl. So, the following data (provided in wlp-lexicon_invalid-sequence.txt) clearly does not satisfy this requirement:

\me jampaly(pa) (N) (PV): (La,Wi,Y)
\gl sharp, pointed \egl
\dm spatial: tactile: \edm
...

The yinarlingi test suite uses the test_code1_ordered() function to catch such cases (notice code1_ok is FALSE on the third line):

library(yinarlingi)

# Get local path to the yinarlingi-provided file, 'wlp-lexicon_invalid-sequence.txt'
invalid_lexicon_path <- system.file("extdata/wlp-lexicon_invalid-sequence.txt", package = "yinarlingi")

test_code1_ordered(invalid_lexicon_path) %>%
    knitr::kable() # For pretty display of table