diff --git a/salient-metaphors-of-anger-in-Indonesian-ms.Rmd b/salient-metaphors-of-anger-in-Indonesian-ms.Rmd index 531ff31..1dcaa4b 100644 --- a/salient-metaphors-of-anger-in-Indonesian-ms.Rmd +++ b/salient-metaphors-of-anger-in-Indonesian-ms.Rmd @@ -41,23 +41,23 @@ source("codes/MARAH-utility-R-functions.R") # Introduction {#intro} -This paper investigates the salient metaphoric and metonymic conceptualisations of the concept [anger]{.smallcaps} in Indonesian ([Glottocode: indo1316](https://glottolog.org/resource/languoid/id/indo1316)) based on a variety of written sources ([§\@ref(method)](#method) for details). Indonesian is the standardised, official variety of Malay used in the Indonesian archipelago. It belongs to "the Malayic subgroup of Western Malayo-Polynesian" [@tadmor_malay-indonesian_2009, 791; @adelaar_dialects_2017, 571] of the Austronesian language. Malay-Indonesian is spoken by almost 280 million speakers combined in Indonesia, Malaysia, and Brunei [@tadmor_malay-indonesian_2009, 791; @adelaar_dialects_2017, 571]. For centuries, Malay-Indonesian has been the "lingua franca" in the regions and developed the colloquial varieties exhibiting great differences among themselves and with the standard language in many aspects of linguistic structures [@tadmor_malay-indonesian_2009, 793]. +This paper investigates the salient metaphoric and metonymic conceptualisations of the concept [anger]{.smallcaps} in standard Indonesian ([Glottocode: indo1316](https://glottolog.org/resource/languoid/id/indo1316)), the official variety of Malay used in the Indonesian archipelago. Indonesian belongs to "the Malayic subgroup of Western Malayo-Polynesian" [@tadmor_malay-indonesian_2009, 791; @adelaar_dialects_2017, 571] of the Austronesian language. Malay-Indonesian is spoken by almost 280 million speakers combined in Indonesia, Malaysia, and Brunei [@tadmor_malay-indonesian_2009, 791; @adelaar_dialects_2017, 571]. It has been the "lingua franca" in the regions and developed the colloquial varieties exhibiting great differences among themselves and with the standard language in many aspects of linguistic structures [@tadmor_malay-indonesian_2009, 793]. ## Previous studies on emotion concepts in Indonesian {#previous-study-emotion} -Heider [-@heider_landscapes_1991] analysed emotion concepts in Minangkabau spoken in West Sumatera, and Indonesian spoken by the Minangkabau and the Javanese communities. The goals are to map the connection between emotion words, and identify prototypical scenarios of the emotions related to the relevant behavioural correlates of the emotions. A recent collection of papers in Fox [-@fox_expressions_2018] offers anthropological linguistic studies of emotions in the regional Austronesian languages of Indonesia. +Heider [-@heider_landscapes_1991] analysed emotion concepts in Minangkabau spoken in West Sumatera, and Indonesian spoken by the Minangkabau and the Javanese communities. The goals were to map the connection between emotion words and identify prototypical scenarios of the emotions related to the relevant behavioural correlates. A recent collection of papers in Fox [-@fox_expressions_2018] offers anthropological linguistic studies of emotions in the regional Austronesian languages of Indonesia. -Shaver et al.'s [-@shaver_structure_2001] psychological study determines the hierarchical and family-resemblance structures of emotion lexicons. They found that American English and Indonesian exhibit similarity in conceptualising [emotion]{.smallcaps} lexicons at (i) the superordinate level (i.e., positive and negative emotions) [@shaver_structure_2001, 215-216], and (ii) the basic-level categories. The Indonesian's prototypical terms for the five basic-level categories refer to the same categories as in American English, namely *cinta* 'love', *senang* 'happiness', *marah* 'anger', *kawatir/takut* 'anxiety/fear', and *sedih* 'sadness' [@shaver_structure_2001, 218]. +Shaver et al.'s [-@shaver_structure_2001] psychological study investigated the hierarchical and family-resemblance structures of emotion lexicons. They found that Indonesian and American English exhibit similarity in conceptualising [emotion]{.smallcaps} at (i) the superordinate level (i.e., positive and negative emotions) [@shaver_structure_2001, 215-216], and (ii) the basic-level categories. The Indonesian's terms for the basic-level categories refer to the same categories as in American English, namely *cinta* 'love', *senang* 'happiness', *marah* 'anger', *kawatir/takut* 'anxiety/fear', and *sedih* 'sadness' [@shaver_structure_2001, 218]. -Linguistic studies in Indonesian/Malay reveal the so-called "psycho-collocations" [@matisoff_hearts_1986], which are lexico-semantic metaphoric expressions referring to various kinds of mental activities and personhood, and which component parts consist of body-part terms. Indonesian/Malay have extensive repertoires of psycho-collocations, especially with the word *hati* 'liver', suggesting the prominence of the liver as the seat of psychological realms in the Malay world [@siahaan_did_2008; @goddard_contrastive_2008; @oey_psycho-collocations_1990; @sather_work_2018; @fox_towards_2018, 12]. Musgrave [-@musgrave_nonsubject_2001] and Mulyadi [-@mulyadi_verba_2012] investigated emotion predicates in Indonesian from the formal syntactic and semantic perspectives. Mulyadi [-@mulyadi_verba_2012] in particular is a contrastive studies on the syntax and semantics of emotion verbs between Indonesian and the Asahan Malay variety spoken in Tanjungbalai (Asahan) in the east coast of North Sumatera, Indonesia. +Linguistic studies in Indonesian/Malay reveal the so-called "psycho-collocations" [@matisoff_hearts_1986], which are lexico-semantic figurative expressions for mental activities and personhood, and which component parts consist of body-part terms. Indonesian/Malay have extensive repertoires of psycho-collocations, especially with the word *hati* 'liver', suggesting the prominence of the liver as the seat of psychological realms in the Malay world [@siahaan_did_2008; @goddard_contrastive_2008; @oey_psycho-collocations_1990; @sather_work_2018; @fox_towards_2018, 12]. Musgrave [-@musgrave_nonsubject_2001] and Mulyadi [-@mulyadi_verba_2012] investigated the syntactic and semantic properties of Indonesian emotion predicates. Mulyadi [-@mulyadi_verba_2012] in particular is a contrastive study of emotion verbs between Indonesian and the Asahan Malay variety spoken in Tanjungbalai (Asahan), North Sumatera, Indonesia. -Indonesian emotions have also been analysed using the *Conceptual Metaphor Theory* (CMT) [@siahaan_did_2008; @siahaan_why_2015; @rajeg_metafora_2013; @rajeg_metaphorical_2019]. Siahaan [-@siahaan_did_2008] examined the cultural conceptualisations of the Indonesian word *hati* 'liver', proposing that (i) liver divination ritual and (ii) ethno-religious belief of *hati* 'liver' as the locus for the living soul underlie the conceptualisations of *hati* 'liver' as the seat of emotion and cognition [see also @goddard_contrastive_2008]. Siahaan's [-@siahaan_why_2015] follow-up study discovered that 'emotion' is the predominant figurative extension of Indonesian temperature terms. Next, Rajeg [-@rajeg_metafora_2013] analysed five Indonesian basic-level emotions. He applied *Configural Frequency Analysis* (CFA) [@gries_statistics_2009] to identify emotion-specific metaphors [@kovecses_metaphor_2000, 35] and examined how the main meaning foci of these metaphors semantically distinguish the emotions. A study by Rajeg [-@rajeg_metaphorical_2019] investigated the distinctive metaphors for [happiness]{.smallcaps} near-synonyms, combining the *MetaNet* (MN) approach [@oana_computational_2017; @petruck_metanet_2016], *Metaphorical Pattern Analysis* (MPA) [@stefanowitsch_happiness_2004; @stefanowitsch_words_2006], and *Multiple Distinctive Collexeme Analysis* (MDCA) [@hilpert_distinctive_2006]; the study reveals that metaphors strongly distinguishing happiness and joy in English [@stefanowitsch_happiness_2004; @stefanowitsch_words_2006] are also those distinguishing the Indonesian equivalences of happiness (i.e., '*kebahagiaan*') and joy ('*kegembiraan*'). +Indonesian emotions have also been analysed using the *Conceptual Metaphor Theory* (CMT) [@siahaan_did_2008; @siahaan_why_2015; @rajeg_metafora_2013; @rajeg_metaphorical_2019]. Siahaan [-@siahaan_did_2008] examined the cultural conceptualisations of *hati* 'liver', proposing that (i) liver divination ritual and (ii) ethno-religious belief of *hati* as the locus for the living soul underlie the conceptualisations of *hati* as the seat of emotion and cognition [see also @goddard_contrastive_2008]. Siahaan's [-@siahaan_why_2015] follow-up study discovered that 'emotion' is the predominant figurative extension of Indonesian temperature terms. Next, Rajeg [-@rajeg_metafora_2013] analysed five Indonesian basic-level emotions. He applied *Configural Frequency Analysis* (CFA) [@gries_statistics_2009] to identify emotion-specific metaphors [@kovecses_metaphor_2000, 35] and examined how these metaphors semantically distinguish the emotions. Another study by Rajeg [-@rajeg_metaphorical_2019] investigated the distinctive metaphors for [happiness]{.smallcaps} near-synonyms, combining the *MetaNet* (MN) approach [@oana_computational_2017; @petruck_metanet_2016], *Metaphorical Pattern Analysis* (MPA) [@stefanowitsch_happiness_2004; @stefanowitsch_words_2006], and *Multiple Distinctive Collexeme Analysis* (MDCA) [@hilpert_distinctive_2006]; the study reveals that metaphors strongly distinguishing happiness and joy in English [@stefanowitsch_happiness_2004; @stefanowitsch_words_2006] are also those distinguishing the Indonesian equivalences of happiness (i.e., '*kebahagiaan*') and joy ('*kegembiraan*'). ## Previous studies on [anger]{.smallcaps} in Indonesian {#previous-study-anger} -Several works have been conducted on Indonesian [anger]{.smallcaps}. Heider [-@heider_landscapes_1991, 57, Table 7] found that, in representing anger, figurative expressions (i.e., *palak* 'stifling; angry' and *panas hati* 'lit. hot liver; angry') received higher rating than the literal expression (i.e. *marah*), suggesting the prominence of the figurative (over the literal) expressions. Heider [-@heider_landscapes_1991, 24-25] also proposed four [anger]{.smallcaps}-like clusters in Minangkabau Indonesian: (i) "anger" clusters (*naik darah* 'lit. rising blood; angry'), (ii) "anger/cruel" clusters (*bengis* 'cruel; harsness'), (iii) "anger/dislike" clusters (*gemas* 'irritated'), and (iv) "anger/trembling" clusters (*gemetar* 'trembling'). The elicited scenarios from the "anger/cruel" clusters revealed that the antecedents of anger "are hurtful acts by others, especially naughty children, and the outcomes are physical violence and verbal abuse" [@heider_landscapes_1991, 80, 116] (e.g., [§\@ref(verbal-behaviour-typebased)](#verbal-behaviour-typebased) and [§\@ref(violent-behaviour-typebased)](#violent-behaviour-typebased)). However, Heider [-@heider_landscapes_1991, 80] noted that in the actual (than elicited) behaviour, Indonesians "mask most anger, and the open expression of anger is strongly disapproved of and negatively sanctioned". The Javanese respondents describe the masking of anger more prominently than the Minangkabau respondents. +Several works have been conducted on Indonesian [anger]{.smallcaps}. Heider [-@heider_landscapes_1991, 57, Table 7] discovered that, in representing anger, figurative expressions (i.e., *palak* 'stifling; angry' and *panas hati* 'lit. hot liver; angry') received higher rating than the literal expression (i.e. *marah*). Heider [-@heider_landscapes_1991, 24-25] also proposed four [anger]{.smallcaps}-like clusters in Minangkabau Indonesian: (i) "anger" clusters (*naik darah* 'lit. rising blood; angry'), (ii) "anger/cruel" clusters (*bengis* 'cruel; harsness'), (iii) "anger/dislike" clusters (*gemas* 'irritated'), and (iv) "anger/trembling" clusters (*gemetar* 'trembling'). The elicited scenarios from the "anger/cruel" clusters revealed that the antecedents of anger "are hurtful acts by others, especially naughty children, and the outcomes are physical violence and verbal abuse" [@heider_landscapes_1991, 80, 116] (e.g., [§\@ref(verbal-behaviour-typebased)](#verbal-behaviour-typebased) and [§\@ref(violent-behaviour-typebased)](#violent-behaviour-typebased)). Heider [-@heider_landscapes_1991, 80] also noted that in the actual (than elicited) behaviour, Indonesians "mask most anger, and the open expression of anger is strongly disapproved of and negatively sanctioned". -Rajeg [-@rajeg_metafora_2013, 211-214] discovered that eight (out of the total 88) metaphors are significantly attracted to *amarah/kemarahan* 'anger'. They are [controlling emotion is controlling a moving object]{.smallcaps}, [emotion is pressurised substance]{.smallcaps}, [emotion is fluid in a container]{.smallcaps}, [emotion is heated fluid in a container]{.smallcaps}, [emotion is fire]{.smallcaps}, ([intensity of]{.smallcaps}) [emotion is temperature]{.smallcaps} ([hot/cold]{.smallcaps}), ([intensity of]{.smallcaps}) [emotion is verticality]{.smallcaps} ([high/low]{.smallcaps}), and [emotion is natural forces]{.smallcaps}. Six metaphors are repelled by [anger]{.smallcaps}: [emotion is a possessable object]{.smallcaps}, [causing emotion is object transfer]{.smallcaps}, [emotion is an accidental motion]{.smallcaps}, [emotion is a journey]{.smallcaps}, [becoming emotion is finding an object]{.smallcaps}, and [emotion is liquid]{.smallcaps}. The statistical attraction of Indonesian [anger]{.smallcaps} to the [heat]{.smallcaps}- and [substance]{.smallcaps}-related metaphors suggests the universality and centrality of these metaphors for [anger]{.smallcaps} as found in different languages [@kovecses_concept_2000], most notably English [@stefanowitsch_words_2006; @holland_cognitive_1987]. Rajeg's [-@rajeg_metafora_2013] quantitative study complements Yuditha's [-@yuditha_indonesian_2013] introspective assessment on the specific metaphors she proposed to be applicable to anger. Lastly, Rajeg [-@rajeg_metaphorical_2014] is a preliminary quantitative investigation on the distinctive metaphorical profiles of five [anger]{.smallcaps} synonyms; it demonstrates that, across the synonyms, Intensity is the prominent aspect highlighted by the metaphorical constructions. +Rajeg [-@rajeg_metafora_2013, 211-214] revealed that eight metaphors are significantly attracted to *amarah/kemarahan* 'anger'. They are [controlling emotion is controlling a moving object]{.smallcaps}, [emotion is pressurised substance]{.smallcaps}, [emotion is fluid in a container]{.smallcaps}, [emotion is heated fluid in a container]{.smallcaps}, [emotion is fire]{.smallcaps}, ([intensity of]{.smallcaps}) [emotion is temperature]{.smallcaps} ([hot/cold]{.smallcaps}), ([intensity of]{.smallcaps}) [emotion is verticality]{.smallcaps} ([high/low]{.smallcaps}), and [emotion is natural forces]{.smallcaps}. Six metaphors are statistically repelled: [emotion is a possessable object]{.smallcaps}, [causing emotion is object transfer]{.smallcaps}, [emotion is an accidental motion]{.smallcaps}, [emotion is a journey]{.smallcaps}, [becoming emotion is finding an object]{.smallcaps}, and [emotion is liquid]{.smallcaps}. The statistical attraction of Indonesian [anger]{.smallcaps} to the [heat]{.smallcaps}- and [substance]{.smallcaps}-related metaphors suggests the universality and centrality of these metaphors for [anger]{.smallcaps} as found in different languages [@kovecses_concept_2000], most notably English [@stefanowitsch_words_2006; @holland_cognitive_1987]. Rajeg's [-@rajeg_metafora_2013] quantitative study complements Yuditha's [-@yuditha_indonesian_2013] introspective assessment on the specific metaphors she proposed to be applicable to anger. Lastly, Rajeg [-@rajeg_metaphorical_2014] is a preliminary quantitative investigation on the distinctive metaphorical profiles of five [anger]{.smallcaps} synonyms; it demonstrates that, across the synonyms, Intensity is the prominent aspect highlighted by the metaphorical constructions. ## Aims {#aims} @@ -119,9 +119,9 @@ root_pvals <- if (root_chisq$p.value < 0.05 & root_chisq$p.value > 0.01) " < 0.0 ``` -The Indonesian terms corresponding to the English "anger" are *marah* 'angry; anger'^[The English translations come from Stevens and Schmidgall-Tellings [-@stevens_comprehensive_2004].], *amarah*^[*Amarah* is the informal variant of *marah* as indicated by the official *Kamus Besar Bahasa Indonesia* (KBBI) *The Big Dictionary of Indonesian* and defined under the entry for *marah*: https://kbbi.kemdikbud.go.id/entri/marah (accessed on October 26, 2021).] 'anger', and *kemarahan* 'fury; rage; anger', the abstract noun derivative from the root *marah*. The choice is based on Shaver et al.'s [-@shaver_structure_2001, 217] finding that *marah* emerges as the prototypical label for [anger]{.smallcaps} category in Indonesian, as it is semantically broader and commonly used in everyday Indonesian. The commonality of *(a)marah* is evident from the frequency data in the **Indonesian Leipzig Corpora** (ILC) [@goldhahn_building_2012]. The combined token frequencies of *(a)marah* (N=`r prettyNum(pull(filter(root, lemma == "(a)marah"), n), big.mark = ",")`) is the highest compared to the other two terms identified by Shaver et al [-@shaver_structure_2001, 218, Table 4], namely *geram* 'furious; angry' (N=`r prettyNum(pull(filter(root, lemma == "geram"), n), big.mark = ",")`) and *berang* 'furious; fury' (N=`r prettyNum(pull(filter(root, lemma == "berang"), n), big.mark = ",")`), as well as to the other near-synonyms of *marah*, namely *murka* 'wrath; anger; fury' (N=`r prettyNum(pull(filter(root, lemma == "murka"), n), big.mark = ",")`) and *gusar* 'angry; offended; annoyed' (N=`r prettyNum(pull(filter(root, lemma == "gusar"), n), big.mark = ",")`) (*X*^2^~goodness-of-fit~=`r format(round(root_chisq$statistic, 1))`; *df*=`r root_chisq$parameter`; *p*~two-tailed~`r root_pvals`). Moreover, the derivative *ke**marah**an* (N=`r prettyNum(pull(filter(ke_an, lemma == "kemarahan"), n), big.mark = ",")`) is also the most frequent compared to the other derivatives based on the other roots, namely *ke**murka**an* (N=`r prettyNum(pull(filter(ke_an, lemma == "kemurkaan"), n), big.mark = ",")`), *ke**gusar**an* (N=`r prettyNum(pull(filter(ke_an, lemma == "kegusaran"), n), big.mark = ",")`), *ke**geram**an* (N=`r prettyNum(pull(filter(ke_an, lemma == "kegeraman"), n), big.mark = ",")`), and *ke**berang**an* (N=`r prettyNum(pull(filter(ke_an, lemma == "keberangan"), n), big.mark = ",")`) (*X*^2^~goodness-of-fit~=`r format(round(ke_an_chisq$statistic, 1))`; *df*=`r ke_an_chisq$parameter`; *p*~two-tailed~`r ke_an_pvals`). +The Indonesian terms corresponding to the English "anger" are *marah* 'angry; anger'^[The English translations come from Stevens and Schmidgall-Tellings [-@stevens_comprehensive_2004].], *amarah*^[*Amarah* is the informal variant of *marah* as indicated by the official *Kamus Besar Bahasa Indonesia* (KBBI) *The Big Dictionary of Indonesian* and defined under the entry for *marah*: https://kbbi.kemdikbud.go.id/entri/marah (accessed on October 26, 2021).] 'anger', and *kemarahan* 'fury; rage; anger', the noun derivative from the root *marah*. The choice is based on Shaver et al.'s [-@shaver_structure_2001, 217] finding that *marah* emerges as the prototypical label for [anger]{.smallcaps} category in Indonesian, as it is semantically broader and commonly used in everyday Indonesian. The commonality of *(a)marah* is evident from the frequency data in the **Indonesian Leipzig Corpora** (ILC) [@goldhahn_building_2012]. The combined token frequencies of *(a)marah* (`r prettyNum(pull(filter(root, lemma == "(a)marah"), n), big.mark = ",")` tokens) is the highest compared to the other terms identified by Shaver et al. [-@shaver_structure_2001, 218, Table 4], namely *geram* 'furious; angry' (`r prettyNum(pull(filter(root, lemma == "geram"), n), big.mark = ",")`) and *berang* 'furious; fury' (`r prettyNum(pull(filter(root, lemma == "berang"), n), big.mark = ",")`), as well as to the other near-synonyms of *marah*, namely *murka* 'wrath; anger; fury' (`r prettyNum(pull(filter(root, lemma == "murka"), n), big.mark = ",")`) and *gusar* 'angry; offended; annoyed' (`r prettyNum(pull(filter(root, lemma == "gusar"), n), big.mark = ",")`) (*X*^2^~goodness-of-fit~=`r format(round(root_chisq$statistic, 1))`; *df*=`r root_chisq$parameter`; *p*~two-tailed~`r root_pvals`). Moreover, the derivative *ke**marah**an* (`r prettyNum(pull(filter(ke_an, lemma == "kemarahan"), n), big.mark = ",")` tokens) is the most frequent compared to other derivatives based on the other roots, namely *ke**murka**an* (`r prettyNum(pull(filter(ke_an, lemma == "kemurkaan"), n), big.mark = ",")`), *ke**gusar**an* (`r prettyNum(pull(filter(ke_an, lemma == "kegusaran"), n), big.mark = ",")`), *ke**geram**an* (`r prettyNum(pull(filter(ke_an, lemma == "kegeraman"), n), big.mark = ",")`), and *ke**berang**an* (`r prettyNum(pull(filter(ke_an, lemma == "keberangan"), n), big.mark = ",")`) (*X*^2^~goodness-of-fit~=`r format(round(ke_an_chisq$statistic, 1))`; *df*=`r ke_an_chisq$parameter`; *p*~two-tailed~`r ke_an_pvals`). -The dataset for the type-based, lexical approach is culled from (i) the Indonesian WordNet (v1.0) [@bond_wordnet_2014]^[The Indonesian WordNet 1.0 is available online at http://compling.hss.ntu.edu.sg/omw/cgi-bin/wn-gridx.cgi?usrname=&gridmode=wnbahasa (last accessed on 22 January 2022.)], (ii) the monolingual *Kamus Bahasa Indonesia* (the Indonesian Dictionary) (KBI) [@kamus_bi_2008], (iii) the official *Tesaurus Tematis Bahasa Indonesia* (the Indonesian Thematic Thesaurus) (http://tesaurus.kemdikbud.go.id/tematis/)^[The thesaurus is compiled and maintained by the Language Development and Cultivation Agency of the Ministy of Education and Culture of the Republic of Indonesia.], and (iv) a bilingual Indonesian-English dictionary [@stevens_comprehensive_2004]. The relevant linguistic expressions were gathered as follows. In the Indonesian sources (WordNet, KBI, and the thesaurus), the three Indonesian [anger]{.smallcaps} words (*marah*, *amarah*, *kemarahan*) were used as the search terms. This approach will retrieve entries/expressions that contain (one of the) [anger]{.smallcaps} terms in their definition. Next, these expressions were checked for the metaphorical and metonymical source domains they evoke by looking at their meanings in the online *Kamus Besar Bahasa Indonesia* (KBBI) (the Big Indonesian Dictionary)^[I used KBI instead of KBBI because KBBI only outputs the brief definition of the three [anger]{.smallcaps} words but not the metaphorical/metonymical expressions related to them, which can be found through the linked thematic thesaurus in the entry. In contrast, the available full PDF version of KBI allows me to do a concordance-like search throughout the PDF using the target terms, which may appear in the definition of an entry related to them. This is not possible in KBBI because the search field is only used for the headword, not the words contained in the definition of the headword. That way, we need to know *a priori* all expressions which definitions contain the [anger]{.smallcaps} words.] following the procedure of the *Metaphor Identification Procedure* (MIP) [@pragglejaz_mip_2007; @pragglejaz_mip_2010] (see below). To gather the data from the bilingual Indonesian-English dictionary, the English terms "anger" and "angry" were entered in the search field of the PDF version of the dictionary; this allows retrieval of the Indonesian expressions which definitions contain the word "angry" or "anger". +The dataset for the type-based, lexical approach is culled from (i) the Indonesian WordNet (v1.0) [@bond_wordnet_2014]^[The Indonesian WordNet 1.0 is available at http://compling.hss.ntu.edu.sg/omw/cgi-bin/wn-gridx.cgi?usrname=&gridmode=wnbahasa (last accessed on 22 January 2022.)], (ii) the monolingual *Kamus Bahasa Indonesia* (the Indonesian Dictionary) (KBI) [@kamus_bi_2008], (iii) the official *Tesaurus Tematis Bahasa Indonesia* (the Indonesian Thematic Thesaurus) (http://tesaurus.kemdikbud.go.id/tematis/)^[The thesaurus is compiled and maintained by the Language Development and Cultivation Agency of the Ministy of Education and Culture of the Republic of Indonesia.], and (iv) a bilingual Indonesian-English dictionary [@stevens_comprehensive_2004]. The relevant linguistic expressions were gathered as follows. From the WordNet, KBI, and the thesaurus, the three Indonesian [anger]{.smallcaps} words (*marah*, *amarah*, *kemarahan*) were used as the search terms. This approach will retrieve expressions that contain (one of) the [anger]{.smallcaps} terms in their definition entries. These expressions were then checked for the evoked metaphorical/metonymical conceptualisations by looking at their meanings in the *Kamus Besar Bahasa Indonesia* (KBBI) (the Big Indonesian Dictionary)^[I used KBI instead of KBBI because KBBI only outputs the brief definition of the three [anger]{.smallcaps} words but not the metaphorical/metonymical expressions related to them, which can be found through the linked thematic thesaurus in the entry. In contrast, the available full PDF version of KBI allows me to do a concordance-like search throughout the PDF using the target terms, which may appear in the definition of an entry related to them. This is not possible in KBBI because the search field is only used for the headword, not the words contained in the definition of the headword. That way, we need to know *a priori* all expressions which definitions contain the [anger]{.smallcaps} words.] following the procedure of the *Metaphor Identification Procedure* (MIP) [@pragglejaz_mip_2007; @pragglejaz_mip_2010] (see below). To gather the data from the bilingual dictionary, the English terms "anger" and "angry" were searched for in the PDF version of the dictionary; this allows retrieval of the Indonesian expressions which English definitions contain the word "angry" or "anger". ```{r leipzig-corpus-information} leipzig_size <- mutate(leipzig_size, @@ -135,17 +135,17 @@ leipzig_size_sources <- leipzig_size %>% ``` -The dataset for the token-based, corpus approach is taken from the corpus files (total size = `r prettyNum(sum(leipzig_size$total_tokens), big.mark = ",")` word-tokens) in the ILC. It is chosen since, to the best of my knowledge, ILC is the only open access source to the largest collection of Indonesian texts^[The alternative is the Indonesian corpus in *Sketch Engine* (SE), which is also from online materials as in ILC. However, SE is a paid service to which the institution I work in does not have paid subscription.] and allows downloading the raw corpus files. ILC mainly consists of randomly chosen websites (`r pull(filter(leipzig_size_sources, sources == "web"), perc_sources)`% of the total size) and online news (`r pull(filter(leipzig_size_sources, sources == "news"), perc_sources)`%), followed by the Wikipedia dumps (`r pull(filter(leipzig_size_sources, sources == "wikipedia"), perc_sources)`%) and a mixture of other sources (`r pull(filter(leipzig_size_sources, sources == "mixed"), perc_sources)`%). +The dataset for the token-based, corpus approach is taken from the corpus files in the ILC (total size = `r prettyNum(sum(leipzig_size$total_tokens), big.mark = ",")` word-tokens). It is chosen since, to the best of my knowledge, ILC is the only open access source to the largest collection of Indonesian texts^[The alternative is the Indonesian corpus in *Sketch Engine* (SE), which is also from online materials as in ILC. However, SE is a paid service to which the institution I work in does not have paid subscription.] and allows downloading the raw corpus files. ILC mainly consists of randomly chosen websites (`r pull(filter(leipzig_size_sources, sources == "web"), perc_sources)`% of the total size) and online news (`r pull(filter(leipzig_size_sources, sources == "news"), perc_sources)`%), followed by the Wikipedia dumps (`r pull(filter(leipzig_size_sources, sources == "wikipedia"), perc_sources)`%) and a mixture of other sources (`r pull(filter(leipzig_size_sources, sources == "mixed"), perc_sources)`%). -Following the *Metaphorical Pattern Analysis* (MPA) [@stefanowitsch_words_2006], one-thousand random concordance lines were retrieved for each *marah*, *amarah*, and *kemarahan*, before manually discarding the irrelevant hits (i.e., duplicates, the predicative and attributive uses of the root *marah*, and the literal uses of the expressions). Next, syntactically relevant collocations of the target terms with the potential source-domain lexical units (LUs) were manually determined [@stefanowitsch_happiness_2004, 138; @sullivan_frames_2013, 3, 5], adopting the *MetaNet* (MN) approach that integrates MPA with Construction Grammar and Frame Semantics [@sullivan_frames_2013; @oana_computational_2017; see also @rajeg_metaphorical_2019 for a recent application in Indonesian]. The MIP procedure [@pragglejaz_mip_2007] was applied to determine whether the collocation of the target terms evoke metaphorical readings. It is determined whether the collocates' contextual meaning when co-occurring with the target-domain terms contrasts with their more basic meaning in other contexts, such that the "contextual meaning can be understood in comparison to the basic meaning" [@rajeg_metaphorical_2019, 64; @pragglejaz_mip_2007, 3; @sullivan_frames_2013, 36]. The KBBI was used to determine the basic meaning of the collocates with reference to MIP's features of basic meaning, namely "more concrete (what they evoke is easier to imagine, see, hear, feel, smell, and taste), related to bodily action, more precise (as opposed to vague), historically older, and are not necessarily the most frequent meanings" [@pragglejaz_mip_2007, 3]. An additional diagnostic to determine the basic meaning is a question proposed by Soriano [-@soriano_conceptualization_2005, 91]: "what exactly each expression 'was literally about'?". To illustrate, consider the following examples showing two different ways to convey the existence of *kemarahan* 'anger'. +As in MPA [@stefanowitsch_words_2006], one-thousand random concordance lines were retrieved for each *marah*, *amarah*, and *kemarahan* from the corpus, before manually discarding the irrelevant hits (i.e., duplicates, the predicative and attributive uses of the root *marah*, and the literal uses of the expressions). Next, syntactically relevant collocations of the target terms with the potential source-domain lexical units (LUs) were manually determined [@stefanowitsch_happiness_2004, 138; @sullivan_frames_2013, 3, 5], adopting the *MetaNet* (MN) approach that integrates MPA with Construction Grammar and Frame Semantics [@sullivan_frames_2013; @oana_computational_2017; see @rajeg_metaphorical_2019 for a recent application to Indonesian]. The MIP [@pragglejaz_mip_2007] was applied to determine whether the collocation of the target terms evoke metaphorical readings. It is determined whether the collocates' contextual meaning, when co-occurring with the [anger]{.smallcaps} terms, contrasts with their more basic meaning in other contexts, such that the "contextual meaning can be understood in comparison to the basic meaning" [@rajeg_metaphorical_2019, 64; @pragglejaz_mip_2007, 3; @sullivan_frames_2013, 36]. The KBBI was used to determine the basic meaning of the collocates with reference to MIP's features of basic meaning, namely "more concrete (what they evoke is easier to imagine, see, hear, feel, smell, and taste), related to bodily action, more precise (as opposed to vague), historically older, and are not necessarily the most frequent meanings" [@pragglejaz_mip_2007, 3]. An additional diagnostic to determine the basic meaning is a question proposed by Soriano [-@soriano_conceptualization_2005, 91]: "what exactly each expression 'was literally about'?". To illustrate, consider the following examples showing two different ways to convey the existence of *kemarahan* 'anger'. (@kemarahan_terjadi) *Kemarahan Presiden Jokowi __terjadi__ saat meninjau Pelabuhan Tanjung Priok (...)* (ind-id_web_2015_3M: 1310067)^[At the end of the numbered example, the source of the example is given in the format "(corpus file name: sentence id)" as in (ind-id_web_2015_3M: 1310067).] (@kemarahan_datang) *(...) kemarahan itu bisa saja __datang__ dalam waktu yang ditentukan (...)* (ind-id_web_2013_1M: 143884). -Example (@kemarahan_terjadi) is considered literal given the verbal collocate *terjadi* 'happen' represents an abstract event as its basic meaning and, in this collocation with *kemarahan* (another abstract domain), *terjadi* is still understood in the abstract domain of [anger]{.smallcaps} [@croft_domain_2003, 192; @sullivan_frames_2013, 9; @rajeg_metaphorical_2019, 99]. In contrast, the collocation of *kemarahan* as the subject of the verb *datang* 'come' (@kemarahan_datang), having a basic meaning in the domain of physical translational motion, induces the metaphorical "domain mapping" [@croft_domain_2003, 192] and interpretation of the verb in the [anger]{.smallcaps} domain [cf. @sullivan_integrating_2016, 147; @dancygier_figurative_2014, 135]. Such metaphoric interpretation of *datang* 'come' emerges due to a mismatch between (i) the semantic type-constraint assigned to the semantic role of the subject of *datang* 'come' (prototypically an animate entity), and (ii) the filler of that role, namely an abstract entity [anger]{.smallcaps} that is literally unable to perform translational motion [cf. @stickles_formalizing_2016, 194; @sullivan_integrating_2016, 148; @brooke-rose_grammar_1958, 1]. The principles of metaphor-domain evocation of certain linguistic elements in a grammatical construction (e.g., the subject-verb or verb-object constructions) have been described extensively by Sullivan [-@sullivan_frames_2013; -@sullivan_integrating_2016]. The unclear cases as to whether the tokens are metaphorical or literal were marked as "?" in the database and can be viewed in the supplementary materials at **https://osf.io/c3p4y/?view_only=50e70ee803fd41ec886547cf33848d49 (CITATION)**. +Example (@kemarahan_terjadi) is considered literal given the verbal collocate *terjadi* 'happen' represents an abstract event as its basic meaning and, in this collocation with *kemarahan* (another abstract domain), *terjadi* is still understood in the abstract domain of [anger]{.smallcaps} [@croft_domain_2003, 192; @sullivan_frames_2013, 9; @rajeg_metaphorical_2019, 99]. In contrast, the collocation of *kemarahan* as the subject of the verb *datang* 'come' (@kemarahan_datang), having a basic meaning in the domain of physical translational motion, induces the metaphorical "domain mapping" [@croft_domain_2003, 192] and interpretation of the verb in the [anger]{.smallcaps} domain [cf. @sullivan_integrating_2016, 147; @dancygier_figurative_2014, 135]. Such metaphoric interpretation of *datang* emerges due to a mismatch between (i) the semantic type-constraint assigned to the semantic role of its subject (prototypically an animate entity), and (ii) the filler of that role, namely an abstract entity [anger]{.smallcaps} that is literally unable to perform a translational motion [cf. @stickles_formalizing_2016, 194; @sullivan_integrating_2016, 148; @brooke-rose_grammar_1958, 1]. The principles of metaphor-domain evocation in a grammatical construction have been described extensively by Sullivan [-@sullivan_frames_2013; -@sullivan_integrating_2016]. The unclear cases as to whether the tokens are metaphorical or literal were marked as "?" in the database and can be viewed in the supplementary materials: **https://osf.io/c3p4y/?view_only=50e70ee803fd41ec886547cf33848d49 (CITATION)**. -The identified metaphorical expressions were then grouped thematically into their metaphorical source domains, adopting the MN approach. MN formalises the central notions in CMT by (i) representing the metaphor-input domains as semantic frames, and (ii) viewing conceptual metaphor (hereafter metaphor) as unidirectional mappings from the source-domain frames to the target-domain frames (including the mappings between the frame roles) that are mediated via the grammatical constructions [@sullivan_frames_2013; @stickles_formalizing_2016; @croft_connecting_2009]. The closest English equivalence of the Indonesian source-domain LUs guides the choice for the source-domain frames in the English MN Wiki repository^[The *MetaNet* main page: https://metaphor.icsi.berkeley.edu/pub/en/index.php/MetaNet_Metaphor_Wiki. The *MetaNet* frame repository: https://metaphor.icsi.berkeley.edu/pub/en/index.php/Category:Frame] [see @lopez_distinguishing_2011 for a similar approach in Spanish]. The thematic grouping also considers the relevant categories from previous studies. The metaphorical mappings within a metaphor are postulated by making use of the frame roles available in a given MN frame entry (when available) or proposed anew based on the semantics of the source-domain LUs. The metonymic source domains are determined via expressions referring to the physiological effects of emotion. +The identified metaphorical expressions were then grouped thematically under their metaphorical source domains, adopting the MN approach. MN formalises the central notions in CMT by (i) representing the metaphor-input domains as semantic frames, and (ii) viewing (conceptual) metaphor as unidirectional mappings from the source-domain frames to the target-domain frames (including the mappings between the frame roles) that are mediated via the grammatical constructions [@sullivan_frames_2013; @stickles_formalizing_2016; @croft_connecting_2009]. The closest English equivalence of the Indonesian source-domain LUs guides the choice for the source-domain frames in the English MN Wiki repository^[The *MetaNet* main page: https://metaphor.icsi.berkeley.edu/pub/en/index.php/MetaNet_Metaphor_Wiki. The *MetaNet* frame repository: https://metaphor.icsi.berkeley.edu/pub/en/index.php/Category:Frame] [see @lopez_distinguishing_2011 for a similar approach in Spanish]. The classification also considers the relevant categories from previous studies. The metaphorical mappings within a metaphor are postulated by making use of the available frame roles in a given MN frame entry or proposed anew based on the semantics of the source-domain LUs. The metonymic source domains are determined via expressions referring to the physiological effects of emotion. The *metaphorical salience* measure for a given metaphor in the token-based, corpus approach takes into account the percentages of: