Linguifex

Module:qlu-pron: Difference between revisions

← Older edit
No edit summary
No edit summary
 
(52 intermediate revisions by the same user not shown)
Line 27: Line 27:
local function dediacv(str, diacritic)
local function dediacv(str, diacritic)
return gsub(mw.ustring.toNFD(str), diacritic, "")
return gsub(mw.ustring.toNFD(str), diacritic, "")
end
local function laxen(v)
local otc = {}
local switch = {["e"] = "i", ["i"] = "ɪ", ["u"] = "ʊ"}
for vc in gmatch(v, ".") do
if switch[vc] then vc = gsub(vc, vc, switch[vc]) end
table.insert(otc, vc)
end
return table.concat(otc)
end
end


Line 63: Line 52:
{"qu?", "ḱ"}, {"þ", "θ"}, {"v", "β"},
{"qu?", "ḱ"}, {"þ", "θ"}, {"v", "β"},
{"%-", ""},
--{"%-", ""},
}
}


Line 89: Line 78:
local phonetic_rules = {
local phonetic_rules = {
-- Gorgia Toscana
-- Gorgia Toscana
{"(" .. vowels .. "·)p(" .. vowels .. ")", "%1ɸ%2"},
{"(" .. vowels .. "·)p(" .. vowels .. ")", "%1ɸ%2"}, {"p$", "ɸ"},
{"(" .. vowels .. "·)b(" .. vowels .. ")", "%1β%2"},
{"(" .. vowels .. "·)b(" .. vowels .. ")", "%1β%2"}, {"b$", "β"},
{"(" .. vowels .. "·)t(" .. vowels .. ")", "%1θ%2"},
{"(" .. vowels .. "·)t(" .. vowels .. ")", "%1θ%2"}, {"t$", "θ"},
{"(" .. vowels .. "·)d(" .. vowels .. ")", "%1ð%2"},
{"(" .. vowels .. "·)d(" .. vowels .. ")", "%1ð%2"}, {"d$", "ð"},
{"(" .. vowels .. "·)k(" .. vowels .. ")", "%1x%2"},
{"(" .. vowels .. "·)k(" .. vowels .. ")", "%1x%2"}, {"k$", "x"},
{"(" .. vowels .. "·)g(" .. vowels .. ")", "%1ɣ%2"},
{"(" .. vowels .. "·)g(" .. vowels .. ")", "%1ɣ%2"}, {"g$", "ɣ"},
{"I", "i"},
{"I", "i"},
{"^ɸ", "f"}, {"(" .. consonants .. ")([·ˈˌ]?)ɸ", "%1%2f"},
{"^ɸ", "f"}, {"(" .. consonants .. ")([·ˈˌ]?)ɸ", "%1%2f"},
{"^β", "v"}, {"(" .. consonants .. ")([·ˈˌ]?)β", "%1%2v"},
{"^β", "v"}, {"(" .. consonants .. ")([·ˈˌ]?)β", "%1%2v"},
{"ḱ", "kʷ"}, {"ǵ", "ɡʷ"},  
{"ḱ", "kʷ"}, {"ǵ", "gʷ"},  
{"k([ieɛj])", "k" .. advanced .. "%1"}, {"g([ieɛj])", "g˖%1"},
{"k([ieɛj])", "k" .. advanced .. "%1"}, {"g([ieɛj])", "g˖%1"},
{"k([oɔu])", "q%1"}, {"g([oɔu])", "ɢ%1"},
{"k([oɔu])", "q%1"}, {"g([oɔu])", "ɢ%1"},
Line 107: Line 96:
-- vowels  
-- vowels  
{"ɔ", "ɔ" .. advanced},
{"iw", "ɪw"},  
{"iw", "ɪw"}, {"uw", "u" .. lowered .. "w"},
{"ew", "e" .. c.diaer .. lowered .. "w"}, {"ow", "o" .. lowered .. "w"},
{"ɛw", "æ" .. c.diaer .. "w"}, {"ɔ" .. advanced .. "w", "ʌw"},
{"ɐw", "ɒw"}, {"aw", "ɑw"},
{"ɐw", "ɒw"}, {"aw", "ɑw"},
{"ij", "ɪj"}, {"uj", "u" .. lowered .. "j"},
{"ij", "ɪj"},
{"ej", "e" .. advanced .. "j"}, {"oj", "o" .. advanced .. "j"},
{"ɛj", "ɛ" .. raised .. "j"}, {"ɔ" .. advanced .. "j", "ɐ" .. raised .. "j"},
-- consonants
{"ʈ", "t"},
{"[ʤʧʃ]", "%1ʷ"},
}
 
local narrow_phonetic_rules = {
{"ɔ", "ɔ" .. advanced}, {"uw", "u" .. lowered .. "w"}, {"ew", "e" .. c.diaer .. lowered .. "w"}, {"ow", "o" .. lowered .. "w"},
{"ɛw", "æ" .. c.diaer .. "w"}, {"ɔ" .. advanced .. "w", "ʌw"}, {"ej", "e" .. advanced .. "j"}, {"oj", "o" .. advanced .. "j"},
{"uj", "u" .. lowered .. "j"}, {"ɛj", "ɛ" .. raised .. "j"}, {"ɔ" .. advanced .. "j", "ɐ" .. raised .. "j"},
{"ɐj", "ɔ" .. raised .. "j"}, {"aj", "a" .. c.diaer .. raised .. "j"},
{"ɐj", "ɔ" .. raised .. "j"}, {"aj", "a" .. c.diaer .. raised .. "j"},
-- consonants
    {"ʦ", "t͡s"}, {"ʣ", "d͡z"},
{"ʈ", "t"}, {"ʦ", "t͡s"}, {"ʣ", "d͡z"},
{"([ntdszl])", "%1" .. laminal}, {"(͡[sz])" .. laminal, "%1" .. dental},  
{"([ntdszl])", "%1" .. laminal}, {"(͡[sz])" .. laminal, "%1" .. dental},  
{"[ʤʧʃ]", "%1ʷ"},
}
}


local upper_rules = {
local upper_rules = {
{"a", "ɐ"}, {"ɸ", "f"}, {"β", "v"}, {"θ", "t"}, {"ð", "d"},
{"a", "ɐ"}, {"ɸ", "f"}, {"β", "v"}, {"θ", "t"}, {"ð", "d"},
{"^[ˌˈ]", ""}, {"[ˌˈ]", "·"}, {"ju", "y"}, {"[ɛI]", "ɐj"}, {"ɔ", "aw"},
{"^[ˌˈ]+", ""}, {"[ˌˈ]+", "·"}, {"··", "·"}, {"ju", "y"}, {"ew", "y"}, {"[ɛI]", "ɐj"}, {"ɔ", "aw"},
{"r", "ɾ"}, {"r·r", "ʀ"}, {"d·([ʣʤ])", "·%1"}, {"t·([ʦʧ])", "·%1"},
{"r", "ɾ"}, {"r·r", "ʀ"}, {"d·([ʣʤ])", "·%1"}, {"t·([ʦʧ])", "·%1"},
{"ʣ", "z"}, {"ʦ", "s"}, {"ʧ", "ʃ"}, {"ʤ", "ʒ"},
{"ʣ", "z"}, {"ʦ", "s"}, {"ʧ", "ʃ"}, {"ʤ", "ʒ"},
Line 133: Line 125:
{"k([ieɛj])", "c%1"}, {"g([ieɛj])", "ɟ%1"},
{"k([ieɛj])", "c%1"}, {"g([ieɛj])", "ɟ%1"},
{"ã", "ɐᵑ"}, {"ẽ", "eᵑ"}, {"ĩ", "iᵑ"}, {"õ", "oᵑ"}, {"ũ", "uᵑ"},  
{"ã", "ɐᵑ"}, {"ẽ", "eᵑ"}, {"ĩ", "iᵑ"}, {"õ", "oᵑ"}, {"ũ", "uᵑ"},  
-- final devoicing
{"b$", "p"}, {"d$", "t"}, {"g$", "k"}, {"β$", "ɸ"}, {"z$", "s"},
}
local bolognese_rules = {
{"ʧ", "ʦ"}, {"ʤ", "ʣ"},
-- apophony
{"e([^·ˈˌ]*[·ˈˌ][^·ˈˌ]*[ijuw]+[^·ˈˌ]*[·ˈˌ])", "i%1"}, {"e([^·ˈˌ]*[·ˈˌ][^·ˈˌ]*[ijuw]+[^·ˈˌ]*)$", "i%1"},
{"ɛ([^·ˈˌ]*[·ˈˌ][^·ˈˌ]*[ijuw]+[^·ˈˌ]*[·ˈˌ])", "e%1"}, {"ɛ([^·ˈˌ]*[·ˈˌ][^·ˈˌ]*[ijuw]+[^·ˈˌ]*)$", "e%1"},
{"o([^·ˈˌ]*[·ˈˌ][^·ˈˌ]*[ijuw]+[^·ˈˌ]*[·ˈˌ])", "u%1"}, {"o([^·ˈˌ]*[·ˈˌ][^·ˈˌ]*[ijuw]+[^·ˈˌ]*)$", "u%1"},
{"ɔ([^·ˈˌ]*[·ˈˌ][^·ˈˌ]*[ijuw]+[^·ˈˌ]*[·ˈˌ])", "o%1"}, {"ɔ([^·ˈˌ]*[·ˈˌ][^·ˈˌ]*[ijuw]+[^·ˈˌ]*)$", "o%1"},
{"([ntdsθzðʦʣlrmŋɲɴ][·ˈˌ]?)[fɸ]", "%1p͡f"}, {"([ntdsθzðʦʣlrmŋɲɴ][·ˈˌ]?)s", "%1ʦ"}, {"([ntdsθzðʦʣlrmŋɲɴ][·ˈˌ]?)θ", "%1t͡θ"},
{"ɣ", "ɣ˕"}, {"ð", "ð" .. lowered}, {"v", "ʋ"},
{"(·[^·]*)ɛ", "%1e"}, {"(·[^·]*)ɔ", "%1o"}, {"^([^·]*)ɛ", "%1e"}, {"^([^·]*)ɔ", "%1o"},
{"(·[^·]*)([eo])", "%1%2" .. lowered}, {"^([^·]*)([eo])", "%1%2" .. lowered},
}
local paulistan_rules = {
{"^([ˈˌ]?)s(" .. consonants .. ")", "es%1%2"}, {"r", "ɾ"}, {"d·([ʣʤ])", "·%1"}, {"t·([ʦʧ])", "·%1"}, {"[ʧʦ]", "s"}, {"ʤ", "ʒ"}, 
{"(" .. consonants .. ")([ˈ·ˌ])%1", "%2%1"}, {"ð", "d"}, {"d([iĩjɪ])", "ʤ%1"}, {"t([iĩjɪ])", "ʧ%1"},
{"^([ˈˌ]?)ɾ", "%1ʁ"}, {"ɾ([ˈˌ·]?)(" .. consonants .. ")", "ɹ%1%2"}, {"([pbtdkg])$", "%1ĭ"},
{"([mnŋɲ])(" .. vowels .. ")([ˈˌ·]?)([mnŋɲ])", "%1%2" .. c.tilde .. "%3%4"},
{"ḱ", "kw"}, {"ǵ", "gw"}, {"(" .. vowels .. ")([ˈˌ·]?)ɲ", "%1" .. c.tilde .. "%2j̃"},
{"(·[^·ˈˌ]*)j", "%1ɪ̯"}, {"^(·[^·ˈˌ]*)j", "%1ɪ̯"}, 
{"(·[^·ˈˌ]*)w", "%1ʊ̯"}, {"^(·[^·ˈˌ]*)w", "%1ʊ̯"}, 
{"(·[^·ˈˌ]*)i", "%1ɪ"}, {"^(·[^·ˈˌ]*)i", "%1ɪ"}, 
{"(·[^·ˈˌ]*)u", "%1ʊ"}, {"^(·[^·ˈˌ]*)u", "%1ʊ"}, 
{"(·[^·ˈˌ]*)e", "%1i"}, {"^(·[^·ˈˌ]*)e", "%1i"},
{"(" .. vowels .. ")([ˈˌ·]?)ŋ", "%1" .. c.tilde .. "%2w" .. c.tilde},
{"ẽ", "ẽj" .. c.tilde}, {"õ", "õw" .. c.tilde}, {c.tilde .. c.tilde, c.tilde},
}
}


Line 138: Line 165:
-- Escaped characters
-- Escaped characters
{"ʤ", "d͡ʒ"}, {"ʧ", "t͡ʃ"}, {"ʈ", "t"}, {"ʦ", "t͡s"}, {"ʣ", "d͡z"},
{"ʤ", "d͡ʒ"}, {"ʧ", "t͡ʃ"}, {"ʈ", "t"}, {"ʦ", "t͡s"}, {"ʣ", "d͡z"},
{"g", "ɡ"}, {"ḱ", "kʷ"}, {"ǵ", "ɡʷ"}, {"I", "i"},
{"g", "g"}, {"ḱ", "kʷ"}, {"ǵ", "gʷ"}, {"I", "i"},
{"·", "."},
{"·?ˈ·?", "ˈ"}, {"·", "."},
}
}


local function syllabify(word)
local function syllabify(word, no_stress)
local clusters = m_table.listToSet({
local clusters = m_table.listToSet({
"s[ptʈkfɸ]", "z[bdgβʤmnlr]",
"s[ptʈkfɸ]", "z[bdgβʤmnlr]",
Line 162: Line 189:
word = gsub(word, "··", "·"); word = gsub(word, "·$", ""); word = gsub(word, "^·", "")
word = gsub(word, "··", "·"); word = gsub(word, "·$", ""); word = gsub(word, "^·", "")
word = gsub(word, "(" .. consonants .. ")·(" .. vowels .. ")", "·%1%2")
word = gsub(word, "(" .. consonants .. ")·(" .. vowels .. ")", "·%1%2")
word = gsub(word, "·?(" .. consonants .. ")%1(" .. vowels .. "*)", "%1·%1%2")
word = gsub(word, "(·?)(" .. consonants .. ")%2(" .. vowels .. "*)", "%2%1%2%3")
word = gsub(word, "·(" .. consonants .. ")$", "%1"); word = gsub(word, "^(" .. consonants .. ")·", "%1")
word = gsub(word, "·(" .. consonants .. ")$", "%1"); word = gsub(word, "^(" .. consonants .. ")·", "%1")
word = gsub(word, "·(.)·", "%1·")
word = gsub(word, "·(.)·", "%1·")
--word = gsub(word, "·s(" .. voiceless .. ")", "s·%1")
word = gsub(word, "·s(" .. voiceless .. ")", "s·%1")
--word = gsub(word, "(" .. consonants .. ")s·(" .. voiceless .. ")", "%1·s%2")
word = gsub(word, "(" .. consonants .. ")s·(" .. voiceless .. ")", "%1·s%2")
word = gsub(word, "ï(" .. vowels .. ")", "i·%1")
word = gsub(word, "ï(" .. vowels .. ")", "i·%1"); word = gsub(word, "ḱ·ḱ", "k·ḱ")
word = gsub(word, "·?d([ʤʣ])", "d·%1"); word = gsub(word, "·tʦ", "t·ʦ")
word = gsub(word, "·?d([ʤʣ])", "d·%1"); word = gsub(word, "·tʦ", "t·ʦ")
word = gsub(word, "·?ŋg", "ŋ·g"); word = gsub(word, "··", "·")
word = gsub(word, "·?ŋg", "ŋ·g"); word = gsub(word, "··", "·")
for cluster, _ in pairs(clusters) do
word = gsub(word, "(" .. consonants .. "+)·(" .. consonants .. "+)", function(c1,c2) return match(c1 .. c2, cluster) and "·" .. c1 .. c2 or c1 .. "·" .. c2 end)
end
word = gsub(word, "%-", "·"); word = gsub(word, "··", "·")
end
end
local syllables = split(word, "·");
local syllables = split(word, "·");
if #syllables == 1 then return table.concat(syllables) end -- account for monosyllables
if no_stress or (#syllables == 1 and not match(word, "[" .. c.acute .. c.circ .. "]")) then
return dediacv(word, "[" .. c.acute .. c.circ .. "]")
end
local first_stress = "[âêîôû]"
local first_stress = "[âêîôû]"
Line 203: Line 238:
end
end


function export.crux(term, outputs)
function export.crux(term, no_stress, outputs)
local ret, dialects = {}, {}
local ret, dialects = {}, {}
term = mw.ustring.lower(term)
term = mw.ustring.lower(term)
Line 211: Line 246:
end
end
term = syllabify(term)
term = syllabify(term, no_stress)
term = term:gsub("·([ˈˌ])", "%1")
term = term:gsub("·?([ˈˌ])·?", "%1")
term = term:gsub("ˌˌ", "ˌ")
term = term:gsub("ˌˌ", "ˌ")
Line 219: Line 254:
phonemic = gsub(phonemic, rule[1], rule[2])
phonemic = gsub(phonemic, rule[1], rule[2])
end
end
local phonetic = phonemic
local phonetic = phonemic
Line 226: Line 260:
end
end
local upper, bolognese, paulistan = phonemic, phonemic, phonemic
local upper, bolognese, paulistan = phonemic, phonetic, phonemic
for _, rule in ipairs(upper_rules) do
for _, rule in ipairs(upper_rules) do
upper = gsub(upper, rule[1], rule[2])
upper = gsub(upper, rule[1], rule[2])
end
end
--[[for _, rule in ipairs(bolognese_rules) do
for _, rule in ipairs(bolognese_rules) do
bolognese = gsub(bolognese, rule[1], rule[2])
bolognese = gsub(bolognese, rule[1], rule[2])
end
end
Line 237: Line 271:
for _, rule in ipairs(paulistan_rules) do
for _, rule in ipairs(paulistan_rules) do
paulistan = gsub(paulistan, rule[1], rule[2])
paulistan = gsub(paulistan, rule[1], rule[2])
end]]
end
for _, rule in ipairs(narrow_phonetic_rules) do
phonetic = gsub(phonetic, rule[1], rule[2])
end
for _, rule in ipairs(last_rules) do
for _, rule in ipairs(last_rules) do
Line 258: Line 296:
end
end


dialects["paulistan"] = paulistan
dialects["bolognese"] = bolognese
dialects["upper"] = upper
dialects["upper"] = upper
dialects["bolognese"] = bolognese
dialects["paulistan"] = paulistan
return {phonemic, phonetic}, dialects
return {phonemic, phonetic}, dialects
Line 287: Line 325:




function separate_word(term)
function separate_word(term, no_stress)
     local phonemic, phonetic, dialects = {}, {}, {}
     local phonemic, phonetic, dialects = {}, {}, {}
local m, t, d = "", "", {}
local m, t, d = "", "", {}
Line 293: Line 331:
if match(term, " ") then
if match(term, " ") then
    for word in gsplit(term, " ") do
    for word in gsplit(term, " ") do
        local ret, d = export.crux(word)
        local ret, d = export.crux(word, no_stress)
        m = ret[1]; t = ret[2]
        m = ret[1]; t = ret[2]
         
         
Line 305: Line 343:
    end
    end
     else
     else
     return export.crux(term)
     return export.crux(term, no_stress)
     end
     end


Line 314: Line 352:
     local parent_args = frame:getParent().args
     local parent_args = frame:getParent().args
     local params = {
     local params = {
         [1] = { default = mw.title.getCurrentTitle().nsText == 'Template' and "aggiu" or mw.title.getCurrentTitle().text },
         [1] = { default = mw.title.getCurrentTitle().nsText == 'Template' and "agghiu" or mw.title.getCurrentTitle().text },
        ["rs"] = {type = 'boolean'},
     }
     }
     local args = require("Module:parameters").process(parent_args, params)
     local args = require("Module:parameters").process(parent_args, params)
     local term = args[1]
     local term = args[1]; local no_stress = args.rs
local ret, dialects = separate_word(term)
local ret, dialects = separate_word(term, no_stress)
     local phonemic = ret[1]; local phonetic = ret[2]
     local phonemic = ret[1]; local phonetic = ret[2]
      
      
     local IPA_args = {{pron = '/' .. (phonemic or "test") .. '/', a = "Standard Ravennese"}, {pron = '[' .. (phonetic or "test2") .. ']'}}
     local IPA_args = {{pron = '/' .. (phonemic or "test") .. '/', a = {"Standard Ravennese"}}, {pron = '[' .. (phonetic or "test2") .. ']'}}
     for key, value in pairs(dialects) do
     for key, value in pairs(dialects) do
     if value ~= phonemic then
     if value ~= phonemic and value ~= phonetic  then
     table.insert(IPA_args, {pron = '[' .. value .. ']', a = {sub(key,1,1)}, separator = "\n** "})
     table.insert(IPA_args, {pron = '[' .. value .. ']', a = {sub(key,1,1)}, separator = "\n** "})
     end
     end
Retrieved from "https://duo.linguifex.com/wiki/Module:qlu-pron"