some_analyzing_scripts.R

library(tidyverse)
library(tidytext)

# get the phonological discription of every language ----------------------
df <- read_csv("andic_dicts.csv")

df |> 
  distinct(glottocode) |> 
  pull(glottocode) |> 
  map(function(x){
    df |> 
      filter(glottocode == x) |> 
      unnest_tokens(sound, ipa, token = stringr::str_split, pattern = "[- ]") |> 
      distinct(sound) |> 
      pull(sound) |> 
      sort()
  }) ->
  sound_list

df |> 
  distinct(glottocode) |> 
  pull(glottocode) ->
  names(sound_list)

sound_list

map(sound_list, function(x){x |> str_detect("['ˌ]", negate = TRUE) |> length()})

map(sound_list, function(i){
  "qχ" %in% i  
})

df |> 
  select(glottocode, reference, ipa) |> 
  unnest_tokens(sound, ipa, token = stringr::str_split, pattern = "[- ]")  |> 
  filter(!is.na(sound)) |> 
  mutate(vowel = ifelse(str_detect(sound, "[ioeau]"), "vowel", "consonant"),
         sound = ifelse(vowel == "vowel", str_remove_all(sound, "['ˌ]"), sound)) |> 
  distinct() |> 
  count(reference, glottocode, vowel) |> 
  pivot_wider(names_from = vowel, values_from = n)

# create a table of correspondences ---------------------------------------
library(tidyverse)
df <- read_csv("andic_dicts.csv")

df |> 
  mutate(languages_source = str_c(language, " (", reference, ")")) |> 
  filter(!is.na(meaning_ru)) |> 
  select(languages_source, ipa, meaning_ru, borrowing_source_language, definition) |> 
  mutate(borrowing_source_language = ifelse(is.na(borrowing_source_language),
                                            "native",
                                            borrowing_source_language)) |> 
  select(languages_source, ipa, meaning_ru, borrowing_source_language) |> 
  group_by(meaning_ru, languages_source) |>
  mutate(id = 1:n()) |> 
  sample_n(1) |> 
  select(-id) |> 
  pivot_longer(names_to = "names", values_to = "value",  c(ipa, borrowing_source_language)) |> 
  mutate(languages_source = ifelse(names != "ipa", str_c(languages_source, " bor"), languages_source)) |>
  select(-names) |> 
  pivot_wider(names_from = languages_source, values_from = value) ->
  result

result$n_languages <- 10-rowSums(is.na(result))/2

result |> 
  filter(n_languages >= 2) |> 
  arrange(-n_languages) |> 
  write_csv("andic_dicts_database_export.csv", na = "")