passive_spacyr.r

library(tidyverse)
library(quanteda)
library(spacyr)

get_passive <- function(text, ratio = F){
  d <- spacy_parse(text, dependency = T)
  
  aux_pass <- d %>% 
    mutate(doc_id = as.numeric(str_remove_all(doc_id, "text"))) %>% 
    group_by(doc_id) %>% 
    summarise(aux_pass = sum(dep_rel=="auxpass"))

  if (ratio==T){
      total_words <- text %>% 
    as_tibble() %>% 
    rowwise %>% 
    mutate(wc = ngram::wordcount(value))
    return(aux_pass$aux_pass/total_words$wc)}
  else{
    return(aux_pass$aux_pass)}
}