The quanteda package provides a framework for quantitative text analysis in R. Its consistent grammar uses corpus_* functions for text management, tokens_* for tokenization, dfm_* for document-feature matrices, textstat_* for statistics, textmodel_* for supervised and unsupervised models, and textplot_* for visualization.

What’s covered#

  • General syntax – object-specific prefixes and companion packages (readtext, spacyr, stopwords)
  • Create a corpus – corpus(), corpus_subset(), corpus_reshape(), corpus_segment(), docvars()
  • Tokenize texts – tokens(), tokens_compound(), tokens_ngrams(), tokens_skipgrams(), tokens_lookup() with dictionaries
  • Extract features – dfm(), dfm_select(), dfm_weight(), dfm_group(), dfm_compress(), fcm()
  • Text statistics – textstat_frequency(), textstat_collocations(), textstat_readability(), kwic()
  • Text models – textmodel_nb() (Naïve Bayes), textmodel_svm(), textmodel_wordscores(), textmodel_wordfish()
  • Visualizations – textplot_wordcloud() and other textplot_* functions