MOE TSM sentences

library(jiebaR)
library(dplyr)
library(readr)
library(stringr)
library(tidyverse)
library(quanteda)
library(tidytext)
library(readxl)
library(officer)
library(flextable)

lemma_tsm <- read_csv("詞目總檔.csv") %>% .$詞目
writeLines(lemma_tsm, 'TSM_dict.txt')
sentence_tsm <- read_csv("例句.csv") %>% 
  select(c(1, 5, 6, 7)) %>%
  rename(編號 = 例句編號, 標音 = 例句標音)

sentence_tsm %>% head(10) %>% knitr::kable(align = 'c')

sentence_tsm %>% nrow
filter(sentence_tsm, str_detect(例句, '有')) %>% nrow

## [1] 14980
## [1] 890

TSM u

+noun
(1) 厝內有人客。(existential)
(2) 我有一本冊。(possessive)
(3) 有人來矣。(presentational)

+verb
(4) 我有食飯矣。(perfective) [+telic, +bounded]
(5) 伊有食菸。(habitual) [-telic, +bounded]
(6) 花有紅。(emphatic) [-telic, -bounded]

# initilize `jiebar`
tagger <- worker(type = "tag",
                 user = "TSM_dict.txt",
                 symbol = T,
                 bylines = F)

segmenter <- worker(user = "TSM_dict.txt",
                    symbol = T,
                    bylines = F)

# define own function
tag_text <- function(x, jiebar){
  segment(x, jiebar) %>%
    paste(names(.), sep = "/", collapse = " ")
}

seg_text <- function(x, jiebar){
  segment(x, jiebar) %>%
    paste(collapse = " ")
}

sentence_tsm_tagged <- sentence_tsm %>%
  mutate(斷詞 = map_chr(例句, seg_text, segmenter),
           斷詞標記 = map_chr(例句, tag_text, tagger))

tsm_u <- sentence_tsm_tagged %>%
  filter(str_detect(斷詞標記, '\\b有/')) 

tsm_u %>% head(10) %>% knitr::kable(align = 'c')

nrow(tsm_u)

## [1] 463

contexts of TSM u

corpus_tsm_u <- corpus(tsm_u,
                    docid_field = '編號',
                    text_field = '斷詞')

kwic(corpus_tsm_u, '有') %>% 
  mutate(context = tsm_u$斷詞標記[match(docname, tsm_u$編號)]) %>%
  mutate(pinyin = tsm_u$標音[match(docname, tsm_u$編號)]) %>%
  select(-from, -to, -pattern) %>%
  head(50) %>% 
  knitr::kable(align = 'c')

# define chunk tokenization function
tokenizer_chunks <- function(input){
  str_split(input, pattern = '(，/x)|(。/x)|(？/x)|(！/x)')
}


# convert text to pattern
tsm_u %>%
  select(編號, 斷詞標記) %>%
  unnest_tokens(chunk, 斷詞標記,
                token = tokenizer_chunks) %>% 
  filter(nzchar(chunk)) %>%
# unnest_tokens(pattern, chunck,
#                token = function(x) str_extract(x, '\\b有/v\\s(\\w+/[^n\\s]+\\s)*?\\w+/(a|v)')) %>%
  head(10) %>% 
  knitr::kable(align = 'c')

write docx files

doc <- read_docx()
for (i in 1:nrow(tsm_u)) {
  doc %>%
    body_add_par(tsm_u[i,]$標音) %>%
    body_add_par(tsm_u[i,]$華語翻譯) %>%
    body_add_par('')
}

df <- qflextable(head(sentence_tsm))
body_add_flextable(doc, df)

print(doc, target = 'TSM_u.docx')

References

鄭良偉. 1992. 台灣話和普通話的動相–時態系統. 中國境內語言暨語言學第一輯: 漢語方言, pp. 179–239.
曹逢甫 & 鄭縈. 1995. 談閩南語 “有” 的五種用法及其間的關係. 中國語文研究, 11, 155-167.

Wu, J. S., & Zheng, Z. R. (2018, May). Toward a unified semantics for Ū in Ū+ situation in Taiwan southern min: A modal-aspectual account. In Workshop on Chinese Lexical Semantics (pp. 408-422). Springer, Cham.

https://davidgohel.github.io/officer/
https://github.com/g0v/moedict-data-twblg