malay_indo_happiness_script_gpwr_imr_2021.Rmd

---
title: "R Notebook with codes for analysis in *Exploring diachronic salience of emotion metaphors*: *A contrastive study of HAPPINESS metaphors in Classical Malay and Indonesian*"
author: '*by* [Gede Primahadi Wijaya Rajeg](https://udayananetworking.unud.ac.id/lecturer/880-gede-primahadi-wijaya-rajeg) <a itemprop="sameAs" content="https://orcid.org/0000-0002-2047-8621" href="https://orcid.org/0000-0002-2047-8621" target="orcid.widget" rel="noopener noreferrer" style="vertical-align:top;"><img src="https://orcid.org/sites/default/files/images/orcid_16x16.png" style="width:1em;margin-right:.5em;" alt="ORCID iD icon"></a> & [I Made Rajeg](https://udayananetworking.unud.ac.id/lecturer/1817-i-made-rajeg) <a itemprop="sameAs" content="https://orcid.org/0000-0001-8989-0203" href="https://orcid.org/0000-0001-8989-0203" target="orcid.widget" rel="noopener noreferrer" style="vertical-align:top;"><img src="https://orcid.org/sites/default/files/images/orcid_16x16.png" style="width:1em;margin-right:.5em;" alt="ORCID iD icon"></a>'
output: 
    html_notebook:
      toc: true
---

```{r setup}
knitr::opts_chunk$set(fig.width = 6, 
                      fig.asp = 0.618, 
                      dpi = 600,
                      echo = TRUE,
                      message = FALSE)

# tidyverse suit of packages is required to run the code
library(tidyverse)

# the gridExtra package is required to run the code
library(gridExtra)

# load the `happyr` package (Rajeg 2019) containing the Indonesian data and R functions for the analyses
library(happyr)
```


```{r load-data, message = FALSE}
# load the MCP root database
malay_root_db <- readr::read_tsv("malay_root_db.txt")

## perform some renaming and data pre-processing on the root database
malay_root_db <- malay_root_db %>%
  mutate(syntax = tolower(syntax),
         syntax = str_replace(syntax, 'unsure', 'widlii'),
         syntax = str_replace(syntax, '^dup$', 'duplicate'),
         period_short = str_replace(period, '(..s$)', '00s'),
         genre = if_else(str_detect(genre, '^let') == TRUE, 'letters_&_documents', genre),
         genre = if_else(str_detect(genre, '^prose') == TRUE, 'prose', genre),
         genre = if_else(str_detect(genre, '^tracts') == TRUE, 'tracts_&_manuals', genre),
         genre = if_else(str_detect(genre, '^news') == TRUE, 'newspapers', genre),
         period = str_replace(period, 's$', ''), 
         period = as.double(period))

# load the MCP nominalised ke- -an forms database
malay_nominalised_db <- readr::read_tsv("malay_nominalised_db.txt")

# load the MCP combined database of root and nominalised ke- -an forms
malay_combined_db <- readr::read_tsv("malay_combined_db.txt")

# load the MCP metaphor data
malaymeta <- readr::read_tsv("malaymeta.txt")
```


# Compute the distribution of the syntax of the root (nominal vs. others)
```{r malay-root-syntax-count}
malay_root_syntax <- malay_root_db %>% 
  count(syntax, words) %>% 
  mutate(syntax_short = if_else(syntax %in% c('nominal'), 'nominal', 'misc.')) %>%
  select(words, syntax, syntax_short, n) %>%
  arrange(words, syntax_short, n)
```

The codes below are for generating Table 2 in the manuscript: *Analyzed tokens for the HAPPINESS nouns in MCP*

```{r malay-token-count-root}
## root tokens in MCP
malay_root_tokens <- malay_root_syntax %>%
  # when activated, the code below will exclude duplicates, proper names, unclear cases)
  # filter(!str_detect(syntax, "dfm|dup|proper|widlii")) %>% 
  group_by(words, syntax_short) %>% 
  summarise(n = sum(n), .groups = "drop") %>% 
  spread(key = syntax_short, value = n, fill = 0) %>% 
  ungroup() %>% 
  mutate(tokens = misc. + nominal) %>% 
  rename(root_others = misc., root_nominal = nominal) %>% 
  select(words, tokens, root_nominal, root_others)
```

```{r malay-token-count-nominalised}
## nominalised tokens in MCP
malay_nominalised_tokens <- malay_nominalised_db %>% 
  count(words) %>% 
  rename(nominalised = n)
```

```{r malay-token-count-combined}
## cross-tab for total tokens of the HAPPINESS concepts in ke-an and root-nominal form
malay_analysed_tokens <- malay_root_tokens %>% 
  full_join(malay_nominalised_tokens, by = "words") %>% 
  replace_na(list(nominalised = 0)) %>% 
  mutate(sum_analysed_nominal = root_nominal + nominalised)

malay_analysed_tokens_tidy <- malay_analysed_tokens %>% 
  select(-tokens, -sum_analysed_nominal) %>% 
  gather(use, n, -words) %>% 
  mutate(words = if_else(use == "nominalised",
                         paste("ke", words, "an", sep = ""),
                         words))

malay_tb_2 <- malay_analysed_tokens %>% 
  select(Concepts = words, 
         `Root tokens` = tokens,
         `Root-nominal` = root_nominal, 
         `*ke*- -*an* tokens` = nominalised, 
         `Total analysed tokens (Root-nominal + *ke*- -*an*)` = sum_analysed_nominal) 
malay_tb_2 %>% knitr::kable(caption = "The analyzed tokens for the HAPPINESS nouns in MCP.")

# uncomment the code below to save the table
# malay_tb_2 %>% write_delim(path = "table_2.txt", delim = "\t")
```

# Genre distribution of the MCP corpus

```{r malay-size-by-genre, message = FALSE, warning = FALSE}
totals_by_text <- readr::read_tsv('totals_by_texts.txt')
totals_by_text1 <- totals_by_text %>% 
  rename(genre = classification) %>% 
  group_by(genre) %>% 
  summarise(n = sum(words), .groups = "drop") %>% 
  mutate(prop = round(n/sum(n) * 100, 2)) %>% 
  arrange(desc(prop))
totals_by_text1
total_mcp <- sum(totals_by_text1$n)
total_mcp
```

# Calculating the normalised frequencies of the HAPPINESS words per periods

Preparing the datasets.

```{r malay-period-freq, message = FALSE, warning = FALSE, eval = FALSE, include = FALSE}
readxl::read_xlsx("Malay Concordance Project_Word totals through time.xlsx", 
                  sheet = "totals_by_periods", 
                  range = "A1:G15") %>% # source: http://mcp.anu.edu.au/Q/words.html#bytime
  select(periods, corp_freq = TOTALS) %>% 
  mutate(periods = gsub("(?<=\\-)([0-9]){2}", "", periods, perl = TRUE)) %>% 
  readr::write_tsv("malay_period_size.txt")
```


```{r malay-period-freq-old-unused, message = FALSE, warning = FALSE, eval = FALSE, include = FALSE}
malay_period_freq_tidy <- readr::read_tsv("malay_period_freq_tidy.txt") %>% 
  group_by(periods) %>% 
  mutate(period_total = sum(n),
         rel_freq = round(n/sum(n), 2),
         perc = round(n/sum(n)*100, 2)) %>% 
  ungroup() %>% 
  mutate(mcp_total = sum(n)) %>%
  mutate(genres = replace(genres, genres=='prose_texts', 'prose'),
         genres = replace(genres, genres=='verse_texts', 'verses')) %>%
  mutate(norm_genre = round((n*1000000)/period_total, 2),
         norm_genre1 = round((n*1000000)/mcp_total, 2)) %>%
  group_by(periods) %>% 
  mutate(norm_period = round((period_total*1000000)/mcp_total, 2)) %>%
  ungroup()
```

```{r happiness-normalised-tokens-per-period, message = FALSE, warning = FALSE}
# corpus freq per periods
corpus_freq <- readr::read_tsv("malay_period_size.txt")

malay_periods_words_genres_freq <- readr::read_tsv("malay_periods_words_genres_freq.txt")
lexemes_freq <- 
  malay_periods_words_genres_freq %>% 
  mutate(lexemes = if_else(form == "nominalised",
                           paste("ke", words, "an", sep = ""),
                           words)) %>% 
  count(period_range, lexemes) %>% 
  rename(periods = period_range)

lexemes_norm_freq <- left_join(lexemes_freq, corpus_freq, 
                               by = "periods") %>% 
  filter(!is.na(lexemes)) %>% 
  mutate(norm_n = round((n/corp_freq) * 1000000, 2))
```


# Generating revised line plot (Figure 1 in the revised manuscript)

```{r line-plot-fig-1, message = FALSE, warning = FALSE}
## REVISED: Line plot for Figure 3 (Figure 1 in the revised manuscript) in the Review of Cognitive Linguistics manuscript (1st submitted draft) =======
# tutorial for combining ggplot2 plots (https://towardsdatascience.com/combining-multiple-ggplot2-plots-for-scientific-publications-7dd9908ebe5c)
lexemes_norm_freq_1 <- lexemes_norm_freq %>% 
  mutate(morphtype = if_else(str_detect(lexemes, "^ke.+an$"), "ke- -an", "root"))

# plot the root first
linep_root <- lexemes_norm_freq_1 %>%
  filter(morphtype == "root") %>% 
  mutate(lexemes = factor(lexemes, 
                          levels = c("senang", "bahagia", "gembira", "riang", "ceria"))) %>% 
  ggplot(aes(x = periods, y = norm_n, colour = lexemes, group = lexemes)) + 
  geom_line(size = 1.5) + 
  theme_bw(base_size = 13) +
  theme(axis.text.x = element_text(angle = 45, 
                                   vjust = 1, 
                                   hjust = 1, 
                                   size = 7.5)) +
  labs(x = NULL,
       subtitle = 'The token frequency of the words in each period is normalized per 1 million words.\nThe normalized frequencies are calculated on the basis of the total word-counts of texts in each period.',
       colour = 'words',
       y = 'frequency per million words') +
  theme(plot.subtitle = element_text(size =8),
        plot.caption = element_text(size = 7),
        axis.title.y = element_text(size = 8),
        legend.title = element_text(size = 8),
        legend.text = element_text(size = 6.5)) +
  annotate(geom = "text", x = 4, y = 260, label = "plot for the root words")

# plot the ke- -an types
linep_nominalised <- lexemes_norm_freq_1 %>%
  filter(morphtype == "ke- -an") %>% 
  mutate(lexemes = factor(lexemes, 
                          levels = c("kesenangan", "kebahagiaan", "kegembiraan", "keriangan"))) %>% 
  ggplot(aes(x = periods, y = norm_n, colour = lexemes, group = lexemes)) + 
  geom_line(size = 1.5) + 
  theme_bw(base_size = 13) +
  theme(axis.text.x = element_text(angle = 45, 
                                   vjust = 1, 
                                   hjust = 1, 
                                   size = 7.5)) +
  labs(x = NULL,
       colour = 'words',
       caption = '\nThe words in the legend are in the descending order of their normalized frequency.',
       y = 'frequency per million words') +
  theme(plot.subtitle = element_text(size =8),
        plot.caption = element_text(size = 7),
        axis.title.y = element_text(size = 8),
        legend.title = element_text(size = 8),
        legend.text = element_text(size = 6.5)) +
  annotate(geom = "text", x = 4, y = 130, label = "plot for the nominalized words")
p_line <- grid.arrange(linep_root, linep_nominalised, nrow = 2)
p_line
# uncomment the code below to save the plot
# ggsave(plot = p_line, "RAJ3-1.png", height = 7, width = 6.5, dpi = 600)
# ggsave(plot = p_line, "Figure1.tiff", height = 7, width = 6.5, dpi = 300, device = "tiff") #accepted manuscript
```



# Analysis for the token and type frequency of the *Malay Concordance Project* (MCP) data

```{r ttr-malay}
ttr_malay <- ttr(malaymeta, "METAPHORS", "LEXICAL_UNIT")
```

Counting the number of conceptual metaphor types in MCP.

```{r metaphor-types-malay}
length(ttr_malay$metaphors)
```

# Analysis for the token and type frequency of the Indonesian data

```{r ttr-indo}
ttr_indo <- ttr(happyr::phd_data_metaphor, "metaphors", "lu")
```

Counting the number of conceptual metaphor types in Indonesian.

```{r metaphor-types-indo}
length(ttr_indo$metaphors)
```

# Print out the table for top-10 metaphors in MCP in terms of their token frequency

```{r ttr-token-freq-malay}
# entrenched metaphor
entrench_malay <- arrange(top_n(ttr_malay, 10, token), desc(token))
entrench_malay_top10 <- select(entrench_malay, metaphors, token, percentage = perc_token)
knitr::kable(mutate(entrench_malay_top10, metaphors = happyr::scaps(metaphors)), caption = "Top-10 metaphors with high token frequency in MCP", row.names = TRUE)
```

# Print out the table for top-10 metaphors in Indonesian in terms of their token frequency

```{r ttr-token-freq-indo}
# entrenched metaphor
entrench_indo <- arrange(top_n(ttr_indo, 10, token), desc(token))
entrench_indo_top10 <- select(entrench_indo, metaphors, token, percentage = perc_token)
knitr::kable(rename(mutate(entrench_indo_top10, metaphors = happyr::scaps(metaphors)), `% of token` = percentage), caption = "Top-10 metaphors with high token frequency in Indonesian", row.names = TRUE)
```

# Combining the top-10 MCP and Indonesian data for high-token frequency metaphors

```{r joint-entrenched, message = FALSE, warning = FALSE}
entrench_indo_top10$lang <- "indonesian"
entrench_malay_top10$lang <- "malay"
entrench_malayindo_top10 <- full_join(entrench_indo_top10,
                                      entrench_malay_top10)
entrench_malayindo_top10 <- mutate(entrench_malayindo_top10,
                                   metaphors = str_replace_all(
                                     metaphors,
                                     "^(intensity of )?happiness is( a(n)?)?", ""))
```

# Plotting Figure 2 in the paper (Section 3.1)

```{r joint-entrenched-plot, fig.asp = 0.65, fig.cap = "Top-10 metaphors with high token frequency in Malay and Indonesian"}
entrench_joint_plot <- entrench_malayindo_top10 %>% 
  ggplot(aes(x = reorder(metaphors, percentage), 
             y = percentage, 
             fill = lang)) + 
  geom_col(position = "dodge") + 
  coord_flip() +
  scale_fill_brewer(name = "Corpus",
                    breaks = c("malay", "indonesian"),
                    labels = c("MCP", "Indonesian")) +
  theme_bw() +
  geom_text(aes(label = token), position = position_dodge(width = 0.9), size = 2.5, hjust = 1.25) +
  labs(x = NULL,
       fill = "corpus",
       y = "% of token",
       subtitle = "Numbers within the bars represent raw token frequency")
entrench_joint_plot
# uncomment the code below to save the plot
# entrench_joint_plot + ggsave("RAJ4.png", height = 4.5, width = 6.5, dpi = 600)
# entrench_joint_plot + ggsave("Figure2.tiff", height = 4.5, width = 6.5, dpi = 300, device = "tiff") # accepted manuscript
```

## Get the information on the (dominant) source frames in the POSSESSABLE OBJECT metaphor

```{r frame-possessable}
poss_frame <- happyr::get_frames("possessable", df = malaymeta, frame_var = "METANET_SOURCE_FRAMES", metaphor_var = "METAPHORS", lexunit_var = "LEXICAL_UNIT")
```

The two prominent frames in terms of their token and type frequencies are GAIN POSSESSION (N = `r poss_frame[poss_frame$frames=="GAIN POSSESSION", ]$n`; Type = `r poss_frame[poss_frame$frames=="GAIN POSSESSION", ]$type`) and TRANSFER SCENARIO (N = `r poss_frame[poss_frame$frames=="TRANSFER SCENARIO", ]$n`; Type = `r poss_frame[poss_frame$frames=="TRANSFER SCENARIO", ]$type`).

```{r frame-desired-goal}
goal_frame_malay <- happyr::get_frames("desired goal$", df = malaymeta, frame_var = "METANET_SOURCE_FRAMES", metaphor_var = "METAPHORS", lexunit_var = "LEXICAL_UNIT")
goal_submet_malay <- happyr::get_submappings("desired", df = malaymeta, metaphor_var = "METAPHORS", submet_var = "SUBMET", lexunit_var = "LEXICAL_UNIT")
goal_frame_indo <- happyr::get_frames("desired goal", df = happyr::phd_data_metaphor)
```

The most predominant MOTION frame is PURSUE (N = `r goal_frame_malay[goal_frame_malay$frames=="PURSUE", ]$n`; Type = `r goal_frame_malay[goal_frame_malay$frames=="PURSUE", ]$type`)


## Codes for discussing the LOCATION-based conceptualizations of HAPPINESS

```{r malay-location-frames}
malay_location_frames <- get_frames("is a location", 
                                    df = malaymeta, 
                                    metaphor_var = "METAPHORS", 
                                    frame_var = "METANET_SOURCE_FRAMES", 
                                    lexunit_var = "LEXICAL_UNIT")

malay_location_frames <- mutate(malay_location_frames, 
                                frames = replace(frames,
                                                 frames == "BOUNDED REGION",
                                                 "BEING IN A BOUNDED REGION"))

malay_location_frames <- malay_location_frames %>% 
  group_by(frames) %>% 
  summarise(n = sum(n), 
            type = sum(type), 
            perc = sum(perc), 
            type_perc = sum(type_perc)) %>% 
  arrange(desc(n))
```

In Malay, frame with highest token frequency motivating the LOCATION mapping is the BOUNDED REGION (N = `r filter(malay_location_frames, n == max(n)) %>% pull(n)`; Type = `r filter(malay_location_frames, n == max(n)) %>% pull(type)`)


# Print out the table for top-10 metaphors in MCP in terms of their type frequency

```{r productive-malay}
prod_malay <- arrange(top_n(ttr_malay, 10, type_lu), desc(type_lu))
prod_malay_top10 <- select(prod_malay, metaphors, token, type = type_lu, percentage_type = perc_type_lu)
knitr::kable(rename(mutate(prod_malay_top10, metaphors = happyr::scaps(metaphors)), `% of type` = percentage_type), caption = "Top-10 metaphors with high type frequency in MCP", row.names = TRUE)
```

Metaphor that now appears in the top-10 list of metaphors with high type frequency in Malay is <span style="font-variant:small-caps;">`r setdiff(prod_malay_top10$metaphors, entrench_malay_top10$metaphors)`</span>. This indicates that top-10 metaphors with high token frequency in Malay also have high type frequency. The rank order of some of these metaphors changes in the type-frequency list. In contrast, rank-ordering metaphors in Indonesian data according to the type frequency brings up `r length(setdiff(prod_indo_top10$metaphors, entrench_indo_top10$metaphors))` metaphors absent from the token frequency list. They are <span style="font-variant:small-caps;">`r setdiff(prod_indo_top10$metaphors, entrench_indo_top10$metaphors)[1]`</span> and <span style="font-variant:small-caps;">`r setdiff(prod_indo_top10$metaphors, entrench_indo_top10$metaphors)[2]`</span> (see [\@ref(tab:productive-indo)](#productive-indo)). 

# Print out the table for top-10 metaphors in Indonesian in terms of their type frequency

```{r productive-indo}
prod_indo <- arrange(top_n(ttr_indo, 10, type_lu), desc(type_lu))
prod_indo_top10 <- select(prod_indo, metaphors, token, type = type_lu, percentage_type = perc_type_lu)
knitr::kable(rename(mutate(prod_indo_top10, metaphors = happyr::scaps(metaphors)), `% of type` = percentage_type), caption = "Top-10 metaphors with high type frequency in Indonesian", row.names = TRUE)
```


# Combining the top-10 MCP and Indonesian data for high-type frequency metaphors

```{r joint-productive, message = FALSE, warning = FALSE}
prod_indo_top10$lang <- "indonesian"
prod_malay_top10$lang <- "malay"
prod_malayindo_top10 <- full_join(prod_indo_top10,
                                      prod_malay_top10)
prod_malayindo_top10 <- mutate(prod_malayindo_top10,
                                   metaphors = str_replace_all(
                                     metaphors,
                                     "^(intensity of )?happiness is( a(n)?)?", ""))
```


```{r joint-productive-plot, fig.asp = 0.65, fig.cap = "Top-10 metaphors with high type frequency in Malay and Indonesian"}
prod_joint_plot <- prod_malayindo_top10 %>% 
  ggplot(aes(x = reorder(metaphors, percentage_type), 
             y = percentage_type, 
             fill = lang)) + 
  geom_col(position = "dodge") + 
  coord_flip() +
  scale_fill_brewer(name = "Corpus",
                    breaks = c("malay", "indonesian"),
                    labels = c("MCP", "Indonesian")) +
  theme_bw() +
  geom_text(aes(label = type), position = position_dodge(width = 0.9), size = 2.5, hjust = 1.25) +
  labs(x = NULL,
       fill = "corpus",
       y = "% of type",
       subtitle = "Numbers within the bars represent raw type frequency")
prod_joint_plot
# uncomment the code below to save the plot
# prod_joint_plot + ggsave("RAJ5.png", height = 4.5, width = 6.5, dpi = 600)
# prod_joint_plot + ggsave("Figure3.tiff", height = 4.5, width = 6.5, dpi = 300, device = "tiff") # accepted manuscript
```



Notes on the PLANT metaphor

`r get_submappings(metaphor = 'plant', df = phd_data_metaphor) %>% filter(type_perc == max(type_perc)) %>% pull(type_perc)`% of the total `r ttr_indo %>% filter(str_detect(metaphors, 'plant')) %>% pull(type_lu)` types of the <span style="font-variant:small-caps;">plant</span> metaphor refers to the growth of the plant. This highlights the existence of <span style="font-variant:small-caps;">happiness</span>.