This project uses a Computational Construction Grammar (CCG) approach and collostructional analysis on a PropBank-annotated corpus. It explores the data gained from this and focuses on the differences between lemmas and their word senses when used in alternating constructions. The R scripts can be run on any .json output given by the CCxG Explorer available here: https://ehai.ai.vub.ac.be/ccxg-explorer/.

The paper itself can be viewed in .qmd (Quarto) or .html. It describes the methodology and results in more detail. For those interested in seeing how the tables and plots were generated, the Quarto files reveal this information. For those who are satisfied with reading about the project without these datails, the HTML file suffices.

To analyse your own .json or reproduce the analysis conducted in this study, follow these steps:

• Install and/or load the necessary packages

  • install.packages(c("base", "graphics", "here", "stats", "utils", "ggplot2", "dplyr", "jsonlite", "stringr", "tibble", "tidyr")
    
    # most of these are available in the tidyverse package
    # install.package("tidyverse")
    
    # load the packages
    library(base)
    library(graphics)
    library(stats)
    library(utils)
    library(ggplot2)
    library(tibble)
    library(jsonlite)
    library(dplyr)
    library(here)
    library(tidyr)
    library(stringr)
    

• Reproducing the analysis
  1. Copy the github repo
  2. Run the script "auto.run.R" located in the "scripts" folder.
  3. The results will be in the "results_analysis/reproduced_analyses" folder.
• Analysing your own data set
  1. Copy the github repo
  2. Open "load_data.R"
  3. Change the filepath to your own file you want to analyse
  4. Save "load.data.R"
  5. Open and run "auto.run.R"

• scripts
  • auto.run.R
    • R Scripts that auto runs a collostructional analysis on a PropBank-annotated corpus and generates the necessary files for analysis.
  • clean_analysis.R
    • R script that cleans the results from the DCA and generates several .csv files that contain the cleaned data analysis.
  • clean_data.R
    • R script that cleans the raw data and creates data frames used for the analysis.
  • coll.analysis.R
    • R script that loads the functions needed for the DCA. Copyright (C) 2022 Stefan Th. Gries (Latest changes in this version: 21 August 2022).
  • load_data.R
    • R script that loads the raw corpus .json file and turns it into a workable data frame.
  • packages_used_scripts.R
    • R script that collects the packages used in the different scripts based on the used functions.
  • run.analysis.auto.R
    • R script that performs a DCA on a set file.
  • run.analysis.fileinput.R
    • R script that performs a DCA on a file choosen by the user.
  • utils.R
    • R script that contains several functions used to interpret and analyze the data.
• data
  • corpus_cleaned
    • contains the several files that can be used for a DCA and MDCA
  • raw_data
    • contains the .json raw data output from the CCxG Explorer
    • contains the loaded raw corpus data
• results_analysis
  • MDCA
    • contains the results from the MDCA
  • ordered_analysis_cxn
    • contains the ordered results from the DCA analysis in full
  • reproduced_analyses
    • contains the results from the reproduced DCA analyses