Module:auto cat/data: Difference between revisions

From Laenkea
Jump to navigation Jump to search
No edit summary
No edit summary
Line 14: Line 14:
-- matched against first.
-- matched against first.
-------------------------
-------------------------
local pos = {
local subtopics = {
    "adjective",
    "adverb",
"article",
    "conjunction",
    "determiner",
    "interjection",
"noun",
    "numeral",
    "particle",
    "participle",
    "phrase",
    "postposition",
    "preposition",
    "pronoun",
    "proper noun",
    "symbol",
    "verb",
}
 
local d_subtopics = {
['biochemistry'] = {'biology','chemistry'},
['biochemistry'] = {'biology','chemistry'},
['enzymes'] = 'biochemistry',
['enzymes'] = 'biochemistry',
['organic chemistry'] = 'chemistry',
['organic chemistry'] = 'chemistry',
['sugars'] = 'chemistry',
['sugars'] = 'chemistry',
['cytology'] = 'biology',
['cytology'] = 'biology',
['genetics'] = 'biology',
['genetics'] = 'biology',
Line 44: Line 26:
['physiology'] = 'biology',
['physiology'] = 'biology',
['taxonomy'] = 'biology',
['taxonomy'] = 'biology',
['phonetics'] = 'linguistics',
['phonetics'] = 'linguistics',
['phonology'] = 'linguistics',
['phonology'] = 'linguistics',
['algebra'] = 'mathematics',
['algebra'] = 'mathematics',
['arithmetic'] = 'mathematics',
['arithmetic'] = 'mathematics',
Line 53: Line 37:
['trigonometry'] = 'mathematics',
['trigonometry'] = 'mathematics',
['trigonometric functions'] = 'trigonometry',
['trigonometric functions'] = 'trigonometry',
['nuclear physics'] = 'physics',
['nuclear physics'] = 'physics',
['months'] = 'time',
['months'] = 'time',
['seasons'] = 'time',
['seasons'] = 'time',
['days'] = 'time',
['days'] = 'time',
}
local pos = {
    "adjective",
    "adverb",
"article",
    "conjunction",
    "determiner",
    "interjection",
"noun",
    "numeral",
    "particle",
    "participle",
    "phrase",
    "postposition",
    "preposition",
    "pronoun",
    "proper noun",
    "symbol",
    "verb",
}
}


Line 416: Line 422:
-- process subtopics
-- process subtopics


for subtopic, supertopics in pairs(d_subtopics) do
for subtopic, supertopics in pairs(subtopics) do
local subcats = {}
local subcats = {}
table.insert(subcats, "%l topics")
table.insert(subcats, "%l topics")

Revision as of 20:35, 4 June 2024

-------------------------
--[[  AUTO CAT DATA  ]]--
-------------------------
-- %l = any language name
-- %u = any language name (different to %l)
-- %c = any language code
-- %p = any part of speech
-- %ps = any part of speech (plural)
-- %s = any string
-------------------------
-- Please note that the order of patterns
-- makes a difference!
-- Patterns higher up in the table will be
-- matched against first.
-------------------------
local subtopics = {
	['biochemistry'] = {'biology','chemistry'},
		['enzymes'] = 'biochemistry',
	
	['organic chemistry'] = 'chemistry',
	['sugars'] = 'chemistry',
	
	['cytology'] = 'biology',
	['genetics'] = 'biology',
	['molecular biology'] = 'biology',
	['physiology'] = 'biology',
	['taxonomy'] = 'biology',
	
	['phonetics'] = 'linguistics',
	['phonology'] = 'linguistics',
	
	['algebra'] = 'mathematics',
	['arithmetic'] = 'mathematics',
	['calculus'] = 'mathematics',
	['geometry'] = 'mathematics',
	['statistics'] = 'mathematics',
	['trigonometry'] = 'mathematics',
		['trigonometric functions'] = 'trigonometry',
	
	['nuclear physics'] = 'physics',
	
	['months'] = 'time',
	['seasons'] = 'time',
	['days'] = 'time',
}

local pos = {
    "adjective",
    "adverb",
	"article",
    "conjunction",
    "determiner",
    "interjection",
	"noun",
    "numeral",
    "particle",
    "participle",
    "phrase",
    "postposition",
    "preposition",
    "pronoun",
    "proper noun",
    "symbol",
    "verb",
}

local cats = {
    -- Languages
    {"All languages", {"Fundamental"}},
    	{"%l language", {"All languages"}},
    		{"%l images", {"%l language"}},
    -- Lexember
    {"Lexember 2023: %l", {"%l language"}},
    	{"Lexember 2023 Day %s: %l", {"Lexember 2023: %l"}},
    -- Rhymes
    {"Rhymes", {"Fundamental"}},
    	{"Rhymes:%l", {"%l language", "Rhymes"}},
    		{"Rhymes:%l/%s", {"Rhymes:%l"}},
    -- Lemmas by language
    {"Lemmas by language", {"Fundamental"}},
    	{"Lemmas subcategories by language", {"Lemmas by language| "}},
    		{"Multiword terms by language", {"Lemmas subcategories by language"}},
    			{"%l multiword terms", {"%l lemmas", "Multiword terms by language"}},
			{"Terms with multiple pronunciations by language", {"Lemmas subcategories by language"}},
    			{"%l terms with multiple pronunciations", {"%l language", "Terms with multiple pronunciations by language"}},
			{"Terms with dialect pronunciations by language", {"Lemmas subcategories by language"}},
    			{"%l terms with %s dialect pronunciation", {"%l language", "Terms with dialect pronunciations by language", "Terms with dialect pronunciations by language"}},
    	{"%ps by language", {"Lemmas by language| "}},
    	{"%l lemmas", {"%l language", "Lemmas by language"}},
    		{"%l %ps", {"%l lemmas", "%ps by language"}},

    -- Non-lemmas by language
    {"Non-lemma forms by language", {"Fundamental"}},
    	{"Non-lemma forms subcategories by language", {"Non-lemma forms by language| "}},
    		{"%p forms by language", {"Non-lemma forms subcategories by language"}},
    	{"%l non-lemma forms", {"%l language", "Non-lemma forms by language"}},
    		{"%l %p forms", {"%l non-lemma forms|%p forms", "%l %ps", "%p forms by language"}},
    
    
    -- Morphemes by language
    {"Morphemes by language", {"Lemmas by language"}},
    {"Affixes by language", {"Morphemes by language"}},
    {"Prefixes by language", {"Affixes by language| "}},
    {"Suffixes by language", {"Affixes by language| "}},
    {"Infixes by language", {"Affixes by language| "}},
    {"Circumfixes by language", {"Affixes by language| "}},
    {"Clitics by language", {"Morphemes by language"}},
    -- Language morphemes
    {"%l morphemes", {"%l lemmas", "Morphemes by language"}},
    {"%l affixes", {"%l morphemes", "%l lemmas", "Affixes by language"}},
    {"%l prefixes", {"%l morphemes", "%l affixes", "Prefixes by language"}},
    	{"%l %s prefixes", {"%l prefixes"}},
    {"%l suffixes", {"%l morphemes", "%l affixes", "Suffixes by language"}},
    	{"Nominal suffixes by language", {"Lemmas subcategories by language"}},
    	{"Verbal suffixes by language", {"Lemmas subcategories by language"}},
    	{"Adjectival suffixes by language", {"Lemmas subcategories by language"}},
    	{"%l nominal suffixes", {"%l suffixes", "Nominal suffixes by language"}},
		{"%l adjectival suffixes", {"%l suffixes", "Adjectival suffixes by language"}},
    	{"%l verbal suffixes", {"%l suffixes", "Verbal suffixes by language"}},
    	{"%l %s suffixes", {"%l suffixes"}},
    {"%l infixes", {"%l morphemes", "%l affixes", "Infixes by language"}},
    	{"%l %s infixes", {"%l infixes"}},
    {"%l circumfixes", {"%l morphemes", "%l affixes", "Circumfixes by language"}},
    {"%l clitics", {"%l morphemes", "Clitics by language"}},
    -- Types of lemma by language
    {"%s %ps by language", {"Lemmas subcategories by language"}},
    {"%ps with multiple genders by language", {"Lemmas subcategories by language"}},
    {"%ps with multiple animacies by language", {"Lemmas subcategories by language"}},
    
    -- Language types of lemma *with cross-linguistic category*
    {"%s %ps by language", {"Lemmas subcategories by language"}},
	    {"%l comparable %ps", {"%l %ps", "Comparable %ps by language"}},
	    {"%l uncomparable %ps", {"%l %ps", "Uncomparable %ps by language"}},
	    {"%l possessive pronouns", {"%l pronouns", "Possessive pronouns by language"}},
	    {"%l imperfective verbs", {"%l verbs", "Imperfective verbs by language"}},
	    {"%l perfective verbs", {"%l verbs", "Perfective verbs by language"}},
	    {"%l biaspectual verbs", {"%l verbs", "Biaspectual verbs by language"}},
	    {"%l transitive verbs", {"%l verbs", "Transitive verbs by language"}},
	    {"%l intransitive verbs", {"%l verbs", "Intransitive verbs by language"}},
	    {"%l auxiliary verbs", {"%l verbs", "Auxiliary verbs by language"}},
	    {"%l copulative verbs", {"%l verbs", "Copulative verbs by language"}},
	    {"%l deponent verbs", {"%l verbs", "Deponent verbs by language"}},
	    {"%l ditransitive verbs", {"%l verbs", "Ditransitive verbs by language"}},
	    {"%l impersonal verbs", {"%l verbs", "Impersonal verbs by language"}},
	    {"%l stative verbs", {"%l verbs", "Stative verbs by language"}},
	    {"%l passive verbs", {"%l verbs", "Passive verbs by language"}},
	    {"%l reflexive verbs", {"%l verbs", "Reflexive verbs by language"}},
	    {"%l mediopassive verbs", {"%l verbs", "Mediopassive verbs by language"}},
	    {"%l neuter %ps", {"%l %ps", "Neuter %ps by language"}},
	    {"%l feminine %ps", {"%l %ps", "Feminine %ps by language"}},
	    {"%l masculine %ps", {"%l %ps", "Masculine %ps by language"}},
	    {"%l common %ps", {"%l %ps", "Common %ps by language"}},
	    {"%l inanimate %ps", {"%l %ps", "Inanimate %ps by language"}},
	    {"%l animate %ps", {"%l %ps", "Animate %ps by language"}},
	    {"%l personal %ps", {"%l %ps", "Personal %ps by language"}},
	    {"%l nonpersonal %ps", {"%l %ps", "Nonpersonal %ps by language"}},
	    {"%l animal %ps", {"%l %ps", "Animal %ps by language"}},
	    {"%l nonvirile %ps", {"%l %ps", "Nonvirile %ps by language"}},
	    {"%l virile %ps", {"%l %ps", "Virile %ps by language"}},
	    
	{"%s terms by language", {"Lemmas subcategories by language"}},
	    {"%l comparable terms", {"%l language", "Comparable terms by language"}},
    	{"%l comparable %ps", {"%l %ps", "Comparable terms by language"}},
	    {"%l uncomparable terms", {"%l language", "Uncomparable terms by language"}},
	    {"%l %s terms", {"%l language", "%s terms by language"}},
		{"%l uncomparable terms", {"%l language", "Uncomparable terms by language"}},
		   
	{"Singularia tantum by language", {"Lemmas subcategories by language"}},
	    {"%l singularia tantum", {"%l nouns", "Singularia tantum by language"}},
	{"Dualia tantum by language", {"Lemmas subcategories by language"}},
	    {"%l dualia tantum", {"%l nouns", "Dualia tantum by language"}},
	{"Pluralia tantum by language", {"Lemmas subcategories by language"}},
	    {"%l pluralia tantum", {"%l nouns", "Pluralia tantum by language"}},
    {"%ps with multiple genders by language", {"Lemmas subcategories by language"}},
	    {"%l %ps with multiple genders", {"%l %ps", "%ps with multiple genders by language"}},
    {"%ps with multiple animacies by language", {"Lemmas subcategories by language"}},
	    {"%l %ps with multiple animacies", {"%l %ps", "%ps with multiple animacies by language"}},
 
	    -- Radestrian-specific non-automatic PoS subcategories
	    {"Radestrian pluralia binaria", {"Radestrian nouns"}},
	    	{"Radestrian pluralia binaria tantum", {"Radestrian nouns", "Radestrian pluralia binaria| "}},
    
    {"%l %s %ps", {"%l %ps"}}, -- sorts all other PoS descriptors under the PoS but not under a cross-linguistic category (e.g. Virile nouns by language). More for language-specific PoS subcategories.
    
    -- Stems
    	{"%l terms by stem type", {"%l language"}},
	    	{"%l %s stems", {"%l terms by stem type"}},
    
    -- Government
    {"%ps governing the %s by language", {"Lemmas subcategories by language"}},
    	{"%l %ps governing the %s", {"%l %ps", "%ps governing the %s by language"}},
    
    -- Roots by language
    {"Root-derived terms by language", {"Fundamental"}},
    	{"Terms derived from %l roots", {"%l roots", "Terms derived from %l| ", "Root-derived terms by language"}},
    		{"Terms derived from the %l root %s", {"Terms derived from %l roots"}},
    {"Terms by root by language", {"Terms by etymology subcategories by language"}},
    {"Roots by language", {"Terms by etymology subcategories by language"}},
    	{"Nominal roots by language", {"Roots by language| "}},
    	{"Numeral roots by language", {"Roots by language| "}},
    	{"Verbal roots by language", {"Roots by language| "}},
    	{"Adjectival roots by language", {"Roots by language| "}},
    	{"Grammatical roots by language", {"Roots by language| "}},
    {"Terms by %l root by language", {"Terms by etymology subcategories by language|%l"}},
    -- Language roots
    {"%l roots", {"%l language", "%l morphemes", "Roots by language"}},
		{"%l nominal roots", {"%l roots", "Nominal roots by language"}},
		{"%l numeral roots", {"%l roots", "Numeral roots by language"}},
		{"%l verbal roots", {"%l roots", "Verbal roots by language"}},
		{"%l adjectival roots", {"%l roots", "Adjectival roots by language"}},
		{"%l grammatical roots", {"%l roots", "Grammatical roots by language"}},
		{"%l terms by root", {"%l terms by etymology", "%l roots", "Terms by root by language"}},
    		{"%l terms belonging to the root %s", {"%l terms by root"}},
    		{"%l terms by %u root", {"%l terms by etymology", "%l terms derived from %u", "Terms by %u root by language"}},
    {"%l terms derived from the %u root %s", {"%l terms by %u root", "Terms derived from the %u root %s"}},
    -- Etymology by language
    {"Terms by etymology by language", {"Fundamental"}},
	    {"Terms by etymology subcategories by language", {"Terms by etymology by language| "}},
	    	-- Language etymology (no source)
		    {"Abbreviations by language", {"Terms by etymology subcategories by language"}},
		    	{"%l abbreviations", {"%l terms by etymology", "Abbreviations by language"}},
		    {"Acronyms by language", {"Terms by etymology subcategories by language"}},
		    	{"%l acronyms", {"%l terms by etymology", "Acronyms by language"}},
		    {"Adjectivisations by language", {"Terms by etymology subcategories by language"}},
    			{"%l adjectivisations", {"%l terms by etymology", "Adjectivisations by language"}},
		    {"Back-formations by language", {"Terms by etymology subcategories by language"}},
		    	{"%l back-formations", {"%l terms by etymology", "Back-formations by language"}},
		    {"Blends by language", {"Terms by etymology subcategories by language"}},
		    	{"%l blends", {"%l terms by etymology", "Blends by language"}},
		    {"Clippings by language", {"Terms by etymology subcategories by language"}},
		    	{"%l clippings", {"%l terms by etymology", "Clippings by language"}},
		    {"Contractions by language", {"Terms by etymology subcategories by language"}},
		    	{"%l contractions", {"%l terms by etymology", "Contractions by language"}},
		    {"Deadjectivals by language", {"Terms by etymology subcategories by language"}},
		    	{"%l deadjectivals", {"%l terms by etymology", "Deadjectivals by language"}},
		    {"Denominals by language", {"Terms by etymology subcategories by language"}},
		    	{"%l denominals", {"%l terms by etymology", "Denominals by language"}},
		    {"Deverbals by language", {"Terms by etymology subcategories by language"}},
		    	{"%l deverbals", {"%l terms by etymology", "Deverbals by language"}},
		    {"Doublets by language", {"Terms by etymology subcategories by language"}},
		    	{"%l doublets", {"%l terms by etymology", "Doublets by language"}},
		    {"Ellipses by language", {"Terms by etymology subcategories by language"}},
		    	{"%l ellipses", {"%l terms by etymology", "Ellipses by language"}},
		    {"Initialisms by language", {"Terms by etymology subcategories by language"}},
		    	{"%l initialisms", {"%l terms by etymology", "Initialisms by language"}},
		    {"Nominalisations by language", {"Terms by etymology subcategories by language"}},
		    	{"%l nominalisations", {"%l terms by etymology", "Nominalisations by language"}},
		    {"Onomatopoeias by language", {"Terms by etymology subcategories by language"}},
		    	{"%l onomatopoeias", {"%l terms by etymology", "Onomatopoeias by language"}},
		    {"Rebracketings by language", {"Terms by etymology subcategories by language"}},
	    	    {"%l rebracketings", {"%l terms by etymology", "Rebracketings by language"}},
		    {"Reduplications by language", {"Terms by etymology subcategories by language"}},
		    	{"%l reduplications", {"%l terms by etymology", "Reduplications by language"}},
		    {"Terms with uncertain etymologies by language", {"Terms by etymology subcategories by language"}},
		    	{"%l terms with uncertain etymologies", {"%l terms by etymology", "Terms with uncertain etymologies by language"}},
		    {"Terms with unknown etymologies by language", {"Terms by etymology subcategories by language"}},
		    	{"%l terms with unknown etymologies", {"%l terms by etymology", "Terms with unknown etymologies by language"}},
		    {"Terms with multiple etymologies by language", {"Terms by etymology subcategories by language"}},
		    	{"%l terms with multiple etymologies", {"%l terms by etymology", "Terms with multiple etymologies by language"}},
		    {"Verbalisations by language", {"Terms by etymology subcategories by language"}},
    			{"%l verbalisations", {"%l terms by etymology", "Verbalisations by language"}},
		{"%l terms by etymology", {"%l language", "Terms by etymology by language"}},

    -- Etymology source by language
    {"Terms derived from %l", {"%l language"}},
    	{"%l terms derived from %u", {"Terms derived from %u"}},
    	{"Terms borrowed from %l", {"Terms derived from %l", "Borrowed terms by language"}},
    {"Inherited terms by language", {"Terms by etymology subcategories by language"}},
    	{"%l inherited terms", {"%l terms by etymology", "Inherited terms by language"}},
    {"Borrowed terms by language", {"Terms by etymology subcategories by language"}},
    	{"%l borrowed terms", {"%l terms by etymology", "Borrowed terms by language"}},
    		{"%l terms borrowed from %s", {"%l borrowed terms", "%l terms derived from %s", "Terms borrowed from %s"}},
    		{"%l orthographic borrowings", {"%l borrowed terms", "Orthographic borrowings by language"}},
    	{"Orthographic borrowings from %s by language", {"Terms borrowed from %s", "Borrowed terms by language"}},
		{"Semantic loans from %s", {"Terms borrowed from %s", "Borrowed terms by language"}},
		{"Phono-semantic matchings from %s", {"Terms borrowed from %s", "Borrowed terms by language"}},
		{"Transliterations of %s", {"Terms borrowed from %s", "Borrowed terms by language"}},
    {"Calques by language", {"Terms by etymology subcategories by language"}},
    	{"%l calques", {"%l terms by etymology", "Calques by language"}},
    		{"%l terms calqued from %s", {"%l calques", "%l terms borrowed from %s", "Terms calqued from %s"}},
    {"Terms inherited from %s", {"Terms derived from %s", "Inherited terms by language"}},
    	{"%l terms inherited from %s", {"%l inherited terms", "%l terms derived from %s", "Terms inherited from %s"}},
    {"Terms calqued from %s", {"Terms borrowed from %s", "Borrowed terms by language"}},
    {"Learned borrowings by language", {"Terms by etymology subcategories by language"}},
    	{"Learned borrowings from %s by language", {"Terms borrowed from %s", "Borrowed terms by language"}},
    			{"%l learned borrowings from %s", {"%l learned borrowings", "%l terms borrowed from %s", "Learned borrowings from %s by language"}},
		{"%l learned borrowings", {"%l borrowed terms", "Learned borrowings by language"}},
    {"Orthographic borrowings by language", {"Terms by etymology subcategories by language"}},
    	{"%l orthographic borrowings from %s", {"%l orthographic borrowings", "%l terms borrowed from %s", "Orthographic borrowings from %s by language"}},
    {"Semantic loans by language", {"Terms by etymology subcategories by language"}},
    	{"%l semantic loans", {"%l borrowed terms", "Semantic loans by language"}},
    		{"%l semantic loans from %s", {"%l semantic loans", "%l terms borrowed from %s", "Semantic loans from %s"}},
    {"Phono-semantic matchings by language", {"Terms by etymology subcategories by language"}},
    	{"%l phono-semantic matchings", {"%l borrowed terms", "Phono-semantic matchings by language"}},
    		{"%l phono-semantic matchings from %s", {"%l phono-semantic matchings", "%l terms borrowed from %s", "Phono-semantic matchings from %s"}},
    {"Transliterations by language", {"Terms by etymology subcategories by language"}},
    	{"%l transliterations", {"%l borrowed terms", "Transliterations by language"}},
    		{"%l transliterations of %s", {"%l transliterations", "%l terms borrowed from %s", "Transliterations of %s"}},

    -- Affix etymology
    {"Compound terms by language", {"Terms by etymology subcategories by language"}},
    	{"%l compound terms", {"%l terms by etymology", "Compound terms by language"}},
    		{"%l %s compounds", {"%l compound terms"}},
    {"Terms by prefix by language", {"Terms by etymology subcategories by language"}},
    	{"%l terms by prefix", {"%l prefixes", "%l terms by etymology", "Terms by prefix by language"}},
    		{"%l terms prefixed with %s", {"%l terms by prefix"}},
    {"Terms by suffix by language", {"Terms by etymology subcategories by language"}},
    	{"%l terms by suffix", {"%l suffixes", "%l terms by etymology", "Terms by suffix by language"}},
    		{"%l terms suffixed with %s", {"%l terms by suffix"}},
    {"Terms by infix by language", {"Terms by etymology subcategories by language"}},
    	{"%l terms by infix", {"%l infixes", "%l terms by etymology", "Terms by infix by language"}},
    		{"%l terms infixed with %s", {"%l terms by infix"}},
    {"Terms by circumfix by language", {"Terms by etymology subcategories by language"}},
    	{"%l terms by circumfix", {"%l circumfixes", "%l terms by etymology", "Terms by circumfix by language"}},
    		{"%l terms circumfixed with %s", {"%l terms by circumfix"}},

    -- Coinage
    {"Coinages by language", {"Terms by etymology subcategories by language"}},
    {"%l coinages", {"%l terms by etymology", "Coinages by language"}},
    
    -- Entry maintenance
    {"Entry maintenance by language", {"Fundamental"}},
    {"Terms with usage examples by language", {"Entry maintenance by language| "}},
    {"%l entry maintenance", {"%l language", "Entry maintenance by language"}},
    {"%l terms with usage examples", {"%l entry maintenance", "Terms with usage examples by language"}},
    
    {"Terms with translations by language", {"Entry maintenance by language| "}},
    	{"%l terms with translations", {"Terms with translations by language"}},
    	{"Terms with %l translations by language", {"Terms with translations by language|%l"}},
    		{"%l terms with %u translations", {"%l terms with translations", "Terms with %u translations by language"}},
    
    -- Miscellaneous terms (custom)
    {"Homophones by language", {"Fundamental"}},
    	{"%l homophones", {"%l language", "Homophones by language"}},
    {"Interlingual homographs", {"Fundamental"}},
    {"Requested terms by language", {"Fundamental"}},
    	{"%l requested terms", {"%l language", "Requested terms by language"}},
    		{"%l terms requested by my %s", {"%l requested terms"}},
    {"Nth terms by language", {"Fundamental"}},
    	{"%l nth terms", {"%l language", "Nth terms by language"}},
    
    -- Usage labels by language
    	-- Language usage labels
    {"Terms by usage by language", {"Fundamental"}},
    	{"%l terms by usage", {"%l language", "Terms by usage by language"}},
    {"Terms by usage subcategories by language", {"Terms by usage by language| "}},
	    {"Archaic terms by language", {"Terms by usage subcategories by language"}},
	    	{"%l archaic terms", {"%l terms by usage", "Archaic terms by language"}},
	    {"Colloquial terms by language", {"Terms by usage subcategories by language"}},
	    	{"%l colloquial terms", {"%l terms by usage", "Colloquial terms by language"}},
	    {"Dated terms by language", {"Terms by usage subcategories by language"}},
	    	{"%l dated terms", {"%l terms by usage", "Dated terms by language"}},
	    {"Derogatory terms by language", {"Terms by usage subcategories by language"}},
	    	{"%l derogatory terms", {"%l terms by usage", "Derogatory terms by language"}},
	    {"Dialectal terms by language", {"Terms by usage subcategories by language"}},
	    	{"%l dialectal terms", {"%l terms by usage", "Dialectal terms by language"}},
	    {"Euphemisms by language", {"Terms by usage subcategories by language"}},
	    	{"%l euphemisms", {"%l terms by usage", "Euphemisms by language"}},
	    {"Formal terms by language", {"Terms by usage subcategories by language"}},
	    	{"%l formal terms", {"%l terms by usage", "Formal terms by language"}},
	    {"Gender-neutral terms by language", {"Terms by usage subcategories by language"}},
	    	{"%l gender-neutral terms", {"%l terms by usage", "Gender-neutral terms by language"}},
	    {"Humble terms by language", {"Terms by usage subcategories by language"}},
	    	{"%l humble terms", {"%l terms by usage", "Humble terms by language"}},
	    {"Jocular terms by language", {"Terms by usage subcategories by language"}},
	    	{"%l jocular terms", {"%l terms by usage", "Jocular terms by language"}},
	    {"Obsolete terms by language", {"Terms by usage subcategories by language"}},
	    	{"%l obsolete terms", {"%l terms by usage", "Obsolete terms by language"}},
	    {"Offensive terms by language", {"Terms by usage subcategories by language"}},
	    	{"%l offensive terms", {"%l terms by usage", "Offensive terms by language"}},
	    {"Pejorative terms by language", {"Terms by usage subcategories by language"}},
	    	{"%l pejorative terms", {"%l terms by usage", "Pejorative terms by language"}},
	    {"Slang terms by language", {"Terms by usage subcategories by language"}},
	    	{"%l slang terms", {"%l terms by usage", "Slang terms by language"}},
	    {"Slurs by language", {"Terms by usage subcategories by language"}},
	    	{"%l slurs", {"%l terms by usage", "Slurs by language"}},
	    {"Vulgar terms by language", {"Terms by usage subcategories by language"}},
			{"%l vulgar terms", {"%l terms by usage", "Vulgar terms by language"}},
    -- Topics
    {"Topics by language", {"Fundamental"}},
    	{"In-world terms by language", {"Topics by language| "}},
    	{"Out-of-world terms by language", {"Topics by language| "}},
    	{"Names by language", {"Topics by language| "}},
    	{"%s topic by language", {"Topics by language| "}},
    	{"%l topics", {"%l language", "Topics by language"}},
    -- Topic subcategories 
    	-- World specificity
	    {"%l in-world terms", {"%l topics", "In-world terms by language"}},
	    {"%l out-of-world terms", {"%l topics", "Out-of-world terms by language"}},
	    -- Names
	    {"%l names", {"%l topics", "Names by language| "}},
	    	{"%l given names", {"%l topics", "Given names by language", "%l names"}},
	    	{"%l surnames", {"%l topics", "Surnames by language", "%l names"}},
    -- Templates
    {"Templates", {"Fundamental"}},
    {"Templates by language", {"Templates"}},
    {"Entry templates", {"Templates"}},
    {"Link templates", {"Templates"}},
    {"Banner templates", {"Templates"}},
    {"Encyclopedia templates", {"Templates"}},
    {"Etymology templates", {"Entry templates"}},
    {"Definition templates", {"Entry templates"}},
    {"Pronunciation templates", {"Entry templates"}},
    {"Translation templates", {"Entry templates"}},
    {"%l templates", {"Templates by language"}},
    -- Modules
    {"Modules", {"Fundamental"}},
    {"Modules by language", {"Modules"}},
    {"%l modules", {"Modules by language"}},
}

-- provides initial capitalisation
local function capitalize(word)
	if mw.ustring.len(word) > 1 then
		return mw.ustring.upper(mw.ustring.sub(word, 1, 1))	.. mw.ustring.sub(word, 2)
	elseif mw.ustring.len(word) == 1 then
		return mw.ustring.upper(word)
	end
	error("Empty input for capitalisation")
end

-- process subtopics

for subtopic, supertopics in pairs(subtopics) do
	local subcats = {}
	table.insert(subcats, "%l topics")
	table.insert(subcats, capitalize(subtopic) .. " topic by language")
	if type(supertopics) == "string" then
		table.insert(subcats, "%c:" .. capitalize(supertopics))
	elseif type(supertopics) == "table" then
		for _, supertopic in ipairs(supertopics) do
			table.insert(subcats, "%c:" .. capitalize(supertopic))
		end
	end
	table.insert(cats, {"%c:" .. capitalize(subtopic), subcats})
	mw.logObject({"%c:" .. capitalize(subtopic), subcats})
end
table.insert(cats, {"%c:%s", {"%l topics", "%s topic by language"}})


return {
	cats = cats,
	pos = pos,
}