Module:fi-pronunciation

---

The following documentation is located at Module:fi-pronunciation/documentation. ^[edit] Categories were auto-generated by Module:module categorization. ^[edit]

Useful links: subpage list • links • transclusions • testcases • sandbox

Implements Template:fi-pronunciation and Template:fi-IPA.

local export = {}

-- used by cartesian_make and cartesian_combine. loaded as necessary
local m_bit32

local m_IPA = require("Module:IPA")
local m_str_utils = require("Module:string utilities")

local find = m_str_utils.find
local gmatch = m_str_utils.gmatch
local gsub = m_str_utils.gsub
local len = m_str_utils.len
local lower = m_str_utils.lower
local match = m_str_utils.match
local sub = m_str_utils.sub
local U = m_str_utils.char

local gsub_lookahead = require("Module:gsub lookahead")

local langcode = "fi"

-- spelled vowels
local vowels = "aeiouyåäö"
-- any of 'vowels' as a pattern
local vowel = "[" .. vowels .. "]"
-- spelled consonants
local consonants = "bcdfghjklmnpqrstvwxzšžʔ"
-- any of 'consonants' as a pattern
local consonant = "[" .. consonants .. "]"
-- straight apostrophe
local apostrophe = "'"
-- "tertiary stress", a weaker secondary stress (either rhythmic or in some compound words). is there a better way to represent this?
local tertiary = "ˌ"
-- include ˣ because final gemination does not affect rhymes
local ipa_symb = "ˣˈˌ"..tertiary.."̯̝̞̠̪" 

-- orthographic symbols that signify separation of syllables
local sep_symbols = "'’./ +\"-"
-- these signify that the next syllable is an "initial" syllable in a word
-- all symbols from here should also be in sep_symbols
local stress_indicators = "ˈˌ" .. tertiary .. "/ +-"
local stress_indicator = "[" .. stress_indicators .. "]"

-- <<SYLLABIFICATION START>>

-- syllabification must also consider * a consonant
local syl_consonants = consonants .. "*"
local syl_consonant = "[" .. syl_consonants .. "]"

-- diphthongs and long vowels
-- 1 = long vowel, 2 = Vi, 3 = VU, 4 = ie/UO
-- 1 and 2 are long vowels/diphthongs everywhere
-- 3 are diphthongs in wordlet-initial syllables or open syllables
-- 4 are diphthongs in wordlet-initial syllables only
local vowel_sequences = {
	["aa"] = 1, ["ee"] = 1, ["ii"] = 1, ["oo"] = 1,
	["uu"] = 1, ["yy"] = 1, ["ää"] = 1, ["öö"] = 1,
	["ai"] = 2, ["ei"] = 2,             ["oi"] = 2,
	["ui"] = 2, ["yi"] = 2, ["äi"] = 2, ["öi"] = 2,
	["au"] = 3, ["eu"] = 3, ["iu"] = 3, ["ou"] = 3,
	["äy"] = 3, ["ey"] = 3, ["iy"] = 3, ["öy"] = 3,
	["ie"] = 4, ["uo"] = 4, ["yö"] = 4
}

-- sonority by consonant (higher number is more sonorous)
local sonorities = {
	["j"] = -1, ["l"] = -1, ["r"] = -1, ["v"] = -1, ["w"] = -1,
	["m"] = -2, ["n"] = -2, ["ŋ"] = -2,
	["h"] = -3, ["z"] = -3, ["ž"] = -3,
	["s"] = -4, ["š"] = -4, ["f"] = -4,
	["b"] = -5, ["d"] = -5, ["g"] = -5,
	["k"] = -6, ["p"] = -6, ["t"] = -6,
}

-- 'native' consonant clusters
local native_consonants = {
	["lkk"] = true, ["nkk"] = true, ["rkk"] = true,
	                ["nks"] = true,
	["lpp"] = true, ["mpp"] = true, ["rpp"] = true,
	["lsk"] = true, ["nsk"] = true, ["rsk"] = true,
	["lss"] = true, ["nss"] = true, ["rss"] = true,
	["lst"] = true, ["nst"] = true, ["rst"] = true,
	["lts"] = true,                 ["rts"] = true,
	["ltt"] = true, ["ntt"] = true, ["rtt"] = true,
}

local function split_by_sonority(cluster)
	local i = len(cluster)
	-- always split two-consonant clusters evenly
	if i <= 2 then
		return i - 1
	end
	
	-- always split 'native' three-consonant clusters to AB.C
	if i == 3 and native_consonants[cluster] then
		return 2
	end

	local ng = find(cluster, "ng")
	-- never split ng
	if ng then
		return ng + 1 + split_by_sonority(sub(cluster, ng + 2))
	end
	
	local max_sonority = 0
	while i > 1 do
		local sonority = sonorities[sub(cluster, i, i)]
		if sonority == nil then return len(cluster) - 1 end
		if sonority >= max_sonority then break end
		max_sonority = sonority
		i = i - 1
	end
	return i
end

-- used for syllabification overrides (any text between these is not syllabified)
local syl_override_start = U(0xEFF0)
local syl_override_end = U(0xEFF1)

local syl_is_consonant = {}
local syl_is_vowel = {}
local syl_is_sep = {}
local syl_is_stressed = {}

for c in gmatch(syl_consonants, ".") do syl_is_consonant[c] = true end
for c in gmatch(vowels, ".") do syl_is_vowel[c] = true end
for c in gmatch(sep_symbols, ".") do syl_is_sep[c] = true end
for c in gmatch(stress_indicators, ".") do syl_is_stressed[c] = true end

-- mode may be one of the following:
--		default: naive hyphenation using Finnish hyphenation rules
--		"dots": use default rules, but add dots between syllable boundaries if allowed by keep_sep_symbols
--		"sonority": split consonant clusters according to the sonority sequencing principle
--				add dots for new splits (i.e. not those incurred by symbols) followed by multiple consonants if keep_sep_symbols allows it.
local function syllabify(word, keep_sep_symbols, mode)
	local result = {} -- output syllables
	local syllable_prefix = nil -- separator symbol if kept
	local syllable_start = 1
	local has_syllable = false
	local cursor = 1
	local found_vowel = false
	local initial_syllable = true -- note: initial in wordlet, not word
	local lower_word = lower(word)

	local function keep_sep(k)
		return (keep_sep_symbols == true or (type(keep_sep_symbols) == "string" and find(keep_sep_symbols, k, 1, true)))
	end

	local dots = ((mode == "dots") and keep_sep(".")) and "." or nil
	local sonority = mode == "sonority"

	local function add_syllable()
		-- add a syllable if not empty
		if has_syllable then
			local syllable_text = sub(word, syllable_start, cursor - 1)
			if syllable_prefix then
				syllable_text = syllable_prefix .. syllable_text
			end
			table.insert(result, syllable_text)

			-- reset state
			syllable_prefix = dots
			initial_syllable = false
			syllable_start = cursor
			has_syllable = false
			found_vowel = false
		end
	end
	
	local len_word = len(word)

	-- prebuild patterns
	local pattern_cc = "^" .. syl_consonant .. "+"
	local pattern_vnc = "^" .. vowel .. "+[^" .. syl_consonants .. "]"
	local pattern_vcv = "^" .. vowel .. "+" .. syl_consonant .. vowel
	local pattern_ccv = pattern_cc .. vowel
	local pattern_sep = "^%(?([" .. sep_symbols .. "])%)?"
	local pattern_stressed = "^%(?[" .. stress_indicators .. "]%)?"

	while cursor <= len_word do
		local symbol = sub(lower_word, cursor, cursor)

		if syl_is_consonant[symbol] then
			-- C...
			local symbol2 = sub(lower_word, cursor + 1, cursor + 1)

			if syl_is_vowel[symbol2] then
				-- CV: end current syllable if we have found a vowel
				if found_vowel then
					add_syllable()
					found_vowel = false
				end
				cursor = cursor + 1
	
			elseif (sonority and found_vowel and syllable_start < cursor
						and syl_is_consonant[symbol2] and match(lower_word, pattern_ccv, cursor)) then
				-- (V)C+V: split by sonority
				local cluster = match(lower_word, pattern_cc, cursor)
				local split_point = split_by_sonority(cluster)
	
				if split_point ~= nil then
					-- split at the specified position.
					cursor = cursor + split_point
					add_syllable()
	
					local len_cluster = len(cluster)
					-- add a dot if dots are allowed, and if the split point is not before the final consonant
					local add_dot = keep_sep(".") and split_point ~= len_cluster - 1
					if add_dot then syllable_prefix = "." end
					cursor = cursor + len_cluster - split_point
					found_vowel = false
				end

			else
				-- C: continue
				cursor = cursor + 1
			end
			has_syllable = true

		elseif syl_is_vowel[symbol] then
			if found_vowel then
				-- already found a vowel, end current syllable
				add_syllable()
			end

			local vowel_length = 1
			found_vowel = true
			
			local vv = sub(lower_word, cursor, cursor + 1)
			local rank = vowel_sequences[vv]
			if rank then
				-- check for diphthongs or long vowels
				local max_rank
				if initial_syllable then
					max_rank = 4
				else
					-- check for an open syllable
					-- the syllable is open if it is followed by a consonant, except if it is followed by a vowel
					local f1 = sub(lower_word, cursor + 2, cursor + 2)
					local f2 = sub(lower_word, cursor + 3, cursor + 3)
					if syl_is_consonant[f1] and not syl_is_vowel[f2] then
						-- closed syllable
						max_rank = 2
					else
						-- open syllable
						max_rank = 3
					end
				end

				if rank <= max_rank then
					vowel_length = 2
				end
			end

			cursor = cursor + vowel_length
			has_syllable = true

		else
			local sepchar
			if symbol == "(" then
				-- look for e.g. (.)
				sepchar = match(lower_word, pattern_sep, cursor)
			else
				sepchar = syl_is_sep[symbol] and symbol
			end

			if sepchar then
				-- separates syllables
				add_syllable()
				
				syllable_prefix = nil
				initial_syllable = syl_is_stressed[sepchar]
				cursor = cursor + (symbol == "(" and 3 or 1)
				if not keep_sep(sepchar) then
					syllable_start = cursor
				end
				found_vowel = false

			elseif symbol == syl_override_start then
				-- override. separate syllables
				if has_syllable then
					add_syllable()
				elseif syllable_start < cursor then
					-- has retained separator character, add to prefix
					syllable_prefix = (syllable_prefix or "") .. sub(lower_word, syllable_start, cursor - 1)
				end

				syllable_start = cursor + (keep_sep(syl_override_start) and 0 or 1)
				local override_end = find(lower_word, syl_override_end, syllable_start)
				if not override_end then error("syl_override_start must end in syl_override_end") end

				has_syllable = true
				if keep_sep(syl_override_end) then
					cursor = override_end + 1
					add_syllable()
				else
					cursor = override_end
					add_syllable()
					cursor = cursor + 1
				end
				syllable_start = cursor

			else
				-- ?: continue
				cursor = cursor + 1
				has_syllable = true
			end
		end
	end

	add_syllable()
	return result
end

export.syllabify = syllabify

-- <<SYLLABIFICATION END>>

-- <<IPA START>>

-- IPA data
-- Finnish vowels in IPA
local ipa_vowels = "ɑeiouyæø"
local ipa_vowel = "[" .. ipa_vowels .. "]"
-- Finnish consonants in IPA
local ipa_consonants = "kptɡgbdfʔsnmŋlrhʋʃʒrjçɦx"
local ipa_consonant = "[" .. ipa_consonants .. "]"
-- IPA diacritics used in transcriptions of Finnish
local ipa_diacritics = "̝̞̠̪"
local ipa_diacritic = "[" .. ipa_diacritics .. "]"
local nonsyllabic = U(0x32F) -- inverted breve below
local unreleased = U(0x31A)
local nasalized = U(0x303)
local long = "ː"

-- IPA: either a spelled consonant/vowel or an IPA consonant/vowel
local spelled_consonant = "[" .. ipa_consonants .. "cšvwxzž]"
local spelled_vowel = "[" .. ipa_vowels .. "aäö]"

local plosives = "kptbdɡ"
local wordlet_sep = "[/-]"

local stress_p = "[ˈˌ" .. tertiary .. "]"
local stress_s = "[ˌ" .. tertiary .. "]"
local stress_pd = "[ˈˌ" .. tertiary .. "%.]"
local stress_sd = "[ˌ" .. tertiary .. "%.]"

local letter_sequence_phonemes = {
	-- ALL FIRST PARTS MUST START WITH ^. longest sequence must come first.
	{ "^ng", "ŋː" },
	{ "^qu", "kʋ" },
	{ "^zz", "ts" },
}

local letters_phonemes = {
	-- symbols outside this list are not changed
	["a"] = "ɑ",
	["ä"] = "æ",
	["ö"] = "ø",
	["å"] = "o",
	
	["c"] = "k",
	["g"] = "ɡ",
	["q"] = "k",
	["v"] = "ʋ",
	["š"] = "ʃ",
	["ž"] = "ʒ",
	
	["x"] = "ks",
	["*"] = "ˣ",
	["'"] = ".",
}

local replacements_narrow = {
	["ɑ"] = "ɑ̝",
	["e"] = "e̞",
	["ø"] = "ø̞",
	["o"] = "o̞",
	["t"] = "t̪",
	["s"] = "s̠"
}

--	This adds letters_phonemes["e"] = "e", letters_phonemes["i"] = "i", etc.
for letter in gmatch("eiouydhfjklmnprstu", ".") do
	letters_phonemes[letter] = letter
end

local function apply_post_fixes(p)
	-- initial <gn> is /gn/
	p = gsub(p, "ˈŋn", "ˈɡn")

	-- ŋ is short before consonant (by default)
	p = gsub(p, "ŋ"..long.."("..ipa_consonant..")", "ŋ%1")

	-- dissimilation of vowels by sandhi
	p = gsub(p, "("..ipa_vowel..ipa_diacritic.."*)(["..long..nonsyllabic.."]?)("..stress_s..")%1", "%1%2%3(ʔ)%1")

	-- C1(.):C2 -> C1(.)(:)C2
	p = gsub(p, "("..ipa_consonant..ipa_diacritic.."*)"..long.."("..ipa_consonant..ipa_diacritic.."*)", "%1("..long..")%2")
	p = gsub(p, "("..ipa_consonant..ipa_diacritic.."*)%.%1("..ipa_consonant..ipa_diacritic.."*)", "%1.(%1)%2")
	p = gsub(p, "("..ipa_consonant..ipa_diacritic.."*)%(%.?%)%1("..ipa_consonant..ipa_diacritic.."*)", "%1(.%1)%2")
	
	return p
end

local function apply_post_fixes_narrow(p)
	-- C1:+C2 -> C1+(C1)C2
	p = gsub(p, "("..ipa_consonant..ipa_diacritic.."*)"..long..tertiary.."("..ipa_consonant..ipa_diacritic.."*)", "%1"..tertiary.."(%1)%2")
	-- C1+C1C2 -> C1+(C1)C2
	-- but C1C2+C2C3 --> C1(C2)+C2C3
	p = gsub(p, "("..ipa_consonant..ipa_diacritic..")("..ipa_consonant..ipa_diacritic.."*)"..tertiary.."(%.?)%2("..ipa_consonant..ipa_diacritic.."*)", "%1(%2)"..tertiary.."%2%3%4")
	p = gsub(p, "("..ipa_consonant..ipa_diacritic.."*)"..tertiary.."(%.?)%1("..ipa_consonant..ipa_diacritic.."*)", "%1"..tertiary.."(%1)%2%3")
	p = gsub(p, "("..ipa_consonant..ipa_diacritic.."*)"..tertiary.."%(?%.?%)%1("..ipa_consonant..ipa_diacritic.."*)", "%1"..tertiary.."(.%1)%2")
	
	-- t is alveolar in /ts/ and /st/
	p = gsub(p, "t̪("..stress_pd.."?%s*)s̠", "t%1s̠")
	p = gsub(p, "s̠("..stress_pd.."?%s*)t̪", "s̠%1t")
	-- n and l become dental in /lt/ /nt/ /tl/ /tn/
	p = gsub(p, "([ln])("..stress_pd.."?%s*)t̪", "%1̪%2t̪")
	p = gsub(p, "t̪("..stress_pd.."?%s*)([ln])", "t̪%1%2̪")

	-- long j, v after i, u diphthong
	p = gsub(p, "(i"..nonsyllabic..")j("..ipa_vowel..")", "%1j("..long..")%2")
	-- /ʋ/ after /u/ usually realized as /w/ (see Suomi, Toivanen and Ylitalo 2008, p. )
	p = gsub(p, "(u"..nonsyllabic..")ʋ("..ipa_vowel..")", "%1w("..long..")%2")
	-- cleanup
	p = gsub(p, "("..stress_s..")%.", "%1")
	p = gsub(p, "("..stress_pd..")"..stress_s, "%1")

    -- tautosyllabic nasals nasalize vowels between them (see Suomi, Toivanen and Ylitalo 2008, p. 22)
	--p = gsub(p, "([mnŋ]"..long.."?)("..ipa_vowel..")("..ipa_diacritic.."*)([mnŋ])(.?)", function (n0, nv, nvd, n1, anchor)
	--	-- this cannot be simplified to "(.?)" => "([^" .. ipa_vowels .. "]?)", otherwise a vowel after would match
	--	if not find(anchor, ipa_vowel) then
	--		return n0 .. nv .. nasalized .. nvd .. n1 .. anchor
	--	end
	--end)
	
	-- sandhi: nm > mm, np > mp, nb > mb, nk > ŋk, ng > ŋg
	p = gsub(p, "nm", "m"..long)
	p = gsub(p, "n("..stress_pd.."?%s*)([ɡk])", "ŋ%1%2")
	p = gsub(p, "n("..stress_pd.."?%s*)([mpb])", "m%1%2")
	p = gsub(p, "[nm]("..stress_pd.."?%s*)([f])", "ɱ%1%2")

	-- handle potentially long consonants over secondary stresses
	p = gsub(p, "("..ipa_consonant..ipa_diacritic.."*)%("..long.."%)("..stress_s..")", "%2%1("..long..")")
	p = gsub(p, "("..stress_s..")("..ipa_consonant..ipa_diacritic.."*)%("..long.."%)", "(%2)%1%2")
	p = gsub(p, "(["..plosives.."]"..ipa_diacritic.."*)("..stress_s..")%1", "%1"..unreleased.."%2%1")
	p = gsub(p, "%((["..plosives.."]"..ipa_diacritic.."*)%)("..stress_s..")%1", "(%1"..unreleased..")%2%1")
	
	-- [k] allophone before front vowels (see Suomi, Toivanen and Ylitalo 2008, p. 27)
	p = gsub(p, "k([eiyæø])", "k̟%1")
	
	return p
end

-- -1 = degenerate
-- 0 = light
-- 1 = heavy with coda
-- 2 = heavy without coda (long vowel/diphthong)
-- 99 = stressed
local class_stressed_syllable = 99
local function classify_syllable(syllable)
	if match(syllable, "^" .. stress_indicator) then
		return class_stressed_syllable
	end
	local cleaned = gsub(syllable, "%(?%*?%)?$", "")
	if match(cleaned, spelled_vowel .. spelled_vowel .. "$") then
		return 2
	elseif match(cleaned, spelled_consonant .. "$") then
		return match(cleaned, spelled_vowel) and 1 or -1
	elseif match(cleaned, spelled_vowel) then
		return 0
	end
	return -1
end

local function has_hiatus(left, right)
	return match(left, spelled_vowel .. "$") and match(right, "^%(?%.?%)?" .. spelled_vowel)
end

local trad_lookahead = 2 -- how many unstressed syllables at most in a single unit, thus max consecutive unstressed syllables
local function add_rhythmic_stress_ltr(hyph, classes)
	-- traditional syllable stress assigning algorithm (e.g. in (V)ISK)
	-- assign from left to right, but shift to later syllables if a light syllable is followed by a heavy syllable

	local add_stress = {}
	local n = #classes
	local prev_stress

	for index, class in ipairs(classes) do
		local stressed = class == class_stressed_syllable
			
		if not stressed and not prev_stress and index < n then
			-- shift stress if current syllable light and a heavy syllable occurs later (except as the last syllable)

			local has_later_heavy_syllable
			local lookahead_end = math.min(index + trad_lookahead, n - 1)
			for i = index + 1, lookahead_end do
				if classes[i] > 0 then
					has_later_heavy_syllable = true
					break
				end
			end

			stressed = index == n - 1 or (class > 0) or not has_later_heavy_syllable
		end
		
		-- check if next syllable already stressed, and do not stress this syllable if so
		if stressed and index < n then
			stressed = stressed and classes[index + 1] ~= class_stressed_syllable
		end

		if index > 1 and stressed then
			add_stress[index] = true
		end
		prev_stress = stressed
	end

	return add_stress
end

local function add_rhythmic_stress_rtl(hyph, classes)
	-- based on Karvonen (2005) "Word prosody in Finnish"
	-- run algorithm for assigning stresses:
	--	(1) analyze each segment started by an already stressed syllable
	--      as a separate prosodic word
	--  (2) assign from *right to left*
	--  (3) final syllable is never stressed (not by this algorithm)
	--  (4) stress may shift back by a single (unstressed) syllable
	--      to a better candidate:
	--			(C)VC is better than (C)V
	--			(C)VV is better than (C)V and (C)VC
	--  (5) if the final syllable in a prosodic word begins with a vowel
	--      and the previous syllable ends in a vowel, skip it

	local add_stress = {}
	local foot = 0
	local word_final = true
	local index = #classes

	while index > 0 do
		local stressed = classes[index] == class_stressed_syllable

		if stressed then
			-- reset
			foot = 0
			word_final = true
		elseif word_final and index > 1 and has_hiatus(hyph[index - 1], hyph[index]) then
			-- hiatus break at the end of a prosodic word: do nothing

		elseif foot == 0 then
			foot = foot + 1

		-- try to assign a stress.
		-- is the previous syllable stressed?
		elseif index > 1 and classes[index - 1] == class_stressed_syllable then
			-- no two stresses in a row, so do nothing.
		else
			-- we will assign a stress somewhere.
			local class = classes[index]
			-- is this syllable preceded by an unstressed syllable
			-- with more priority?
			if index > 2 and classes[index - 2] ~= class_stressed_syllable and 
							classes[index - 1] > classes[index] then
				-- shift back by one syllable
				index = index - 1
			end

			-- assign stress, and continue.
			add_stress[index] = true
			foot = 0
		end
		index = index - 1
		word_final = false
	end

	return add_stress
end

local function add_rhythmic_stress_collect(hyph, classes, add_stress)
	-- add stress marks
	local rhythmic_stress = tertiary
	local collected = {}
	for index = 1, #hyph do
		if add_stress[index] and classes[index] < class_stressed_syllable then
			local syllable = hyph[index]
			if index > 1 and match(syllable, "^g") and match(hyph[index - 1], "[nŋ]$") then
				-- make sure [nŋ].g > ŋ.ŋ
				collected[index - 1] = gsub(collected[index - 1], "n$", "ŋ")
				syllable = gsub(syllable, "^g", "ŋ")
			end
			collected[index] = rhythmic_stress .. gsub(syllable, "^%.", "")
		else
			collected[index] = hyph[index]
		end
	end
	return table.concat(collected)
end

-- applied *before* IPA conversion
-- returns one result, or possibly two results
local function add_rhythmic_stress(word, assign_rtl)
	-- keep_sep_symbols = true
	local hyph = syllabify(word, true, "sonority")
	local final_heavy = false
	local collected
	local add_rhythmic_stress_impl = assign_rtl and add_rhythmic_stress_rtl or add_rhythmic_stress_ltr

	-- find stressed syllables and classify all syllables
	local classes = {}
	for index, syllable in ipairs(hyph) do
		if index == 1 then
			classes[index] = class_stressed_syllable
		else
			classes[index] = classify_syllable(syllable)
		end
	end

	local add_stress
	-- final heavy syllable may be stressed if preceded by an unstressed
	-- light syllable (in words with at least 3 syllables)
	final_heavy = #classes >= 3 and classes[#classes] > 0 and classes[#classes - 1] == 0

	add_stress = add_rhythmic_stress_impl(hyph, classes)

	local res1 = add_rhythmic_stress_collect(hyph, classes, add_stress)
	local res2

	if final_heavy then
		-- make final syllable stressed
		local actual_final = classes[#classes]
		classes[#classes] = class_stressed_syllable
		add_stress = add_rhythmic_stress_impl(hyph, classes)
		-- so that add_rhythmic_stress_collect actually adds it
		classes[#classes] = actual_final
		add_stress[#classes] = true
		res2 = add_rhythmic_stress_collect(hyph, classes, add_stress)
	end

	return res1, res2
end

local pattern_syllable_c = "^" .. consonant
local pattern_syllable_open = "^" .. consonant .. vowel

local function IPA_handle_diphthongs_replace_open_noninitial(diphthong, after1, after2)
	if syl_is_consonant[after1] then
		-- consonant after diphthong
		-- must be followed by vowel so that it's part of the
		-- following syllable, else it's in this syllable
		-- and thus this syllabie is closed

		if syl_is_vowel[after2] then
			-- single consonant followed by vowel -> this syllable is open
			return diphthong .. nonsyllabic, after1 .. after2
		else
			-- single consonant not followed by vowel -> this syllable is closed
			return diphthong, after1 .. after2
		end
	end
	if after1 == nonsyllabic then
		-- already marked as nonsyllabic
		return diphthong, after1 .. after2
	end	
	-- no consonant after diphthong -> this syllable is open
	return diphthong .. nonsyllabic, after1 .. after2
end

local pattern_word_initial = "^[^" .. stress_indicators .. vowels .. "]*"
local pattern_after_sep = stress_indicator .. "[^" .. vowels .. "]*"

local function IPA_handle_diphthongs_word_initial(IPA, diphthong)
	IPA = gsub(IPA, pattern_word_initial .. diphthong, "%0" .. nonsyllabic)
	IPA = gsub(IPA, pattern_after_sep .. diphthong, "%0" .. nonsyllabic)
	return IPA
end

local function IPA_handle_diphthongs_word_initial_or_open(IPA, diphthong)
	IPA = gsub(IPA, pattern_word_initial .. diphthong, "%0" .. nonsyllabic)
	IPA = gsub(IPA, pattern_after_sep .. diphthong, "%0" .. nonsyllabic)
	IPA = gsub_lookahead(IPA, "(" .. diphthong .. ")(.?)(.?)", IPA_handle_diphthongs_replace_open_noninitial)
	return IPA
end

-- Add nonsyllabic diacritic after last vowel of diphthong.
local function IPA_handle_diphthongs(IPA)
	-- /Vi/ everywhere
	IPA = gsub(IPA, "[aeouyäö]i", "%0" .. nonsyllabic)

	-- /ie/, /uo/, /yö/ root-initially
	IPA = IPA_handle_diphthongs_word_initial(IPA, "ie")
	IPA = IPA_handle_diphthongs_word_initial(IPA, "uo")
	IPA = IPA_handle_diphthongs_word_initial(IPA, "yö")

	-- /VU/ root-initially or in open syllables
	IPA = IPA_handle_diphthongs_word_initial_or_open(IPA, "[aoei]u")
	IPA = IPA_handle_diphthongs_word_initial_or_open(IPA, "[äöei]y")
	
	return IPA
end

local function fi_IPA_wordlet(term, is_narrow)
	local cursor = 1
	local len_term = len(term)
	local phonemes = {}
	
	while cursor <= len_term do
		local target
		
		-- try letter_sequence_phonemes first
		for _, replacement in ipairs(letter_sequence_phonemes) do
			if match(term, replacement[1], cursor) then
				target = replacement[2]
				cursor = cursor + len(replacement[1]) - 1   -- account for ^
				break
			end
		end

		-- then letters_phonemes
		if not target then
			local letter = sub(term, cursor, cursor)
			target = letters_phonemes[letter] or letter
			cursor = cursor + 1
		end

		table.insert(phonemes, target or letter)
	end
	
	local result = table.concat(phonemes)
	result = gsub(result, "nk", "ŋk")
	result = gsub(result, "ŋ[ŋ" .. long .. "]$", "ŋ")
	
	if is_narrow then
		-- articulation of h (Suomi, Toivanen & Ylitalo 2008, p. 28)
		result = gsub_lookahead(result, "(.?)h(.?)",
			function (before, after)
				local h
				if after ~= "" and after ~= "h" then
					if before ~= "" and (ipa_vowels:find(before) or before == nonsyllabic) then
						if ipa_consonants:find(after) then
							-- vihma, yhtiö
							if before == "i" or before == "y" then
								h = "ç"
							-- mahti, kohme, tuhka
							elseif before == "ɑ" or before == "o" or before == "u" then
								h = "x"
							end
						-- maha
						elseif ipa_vowels:find(after) then
							h = "ɦ"
						end
					end
				end
				
				if h then
					return before .. h, after
				end
			end)
		
		-- double letter replacement must be handled earlier here
		result = gsub(result, "(%a)%1", "%1" .. long)
	
		for letter, phoneme in pairs(replacements_narrow) do
			result = gsub(result, letter, phoneme)
		end
	end
	
	return result
end

local function fi_IPA_termwordlets(term, is_narrow, hyphenstress)
	local wordlet_start = 1
	local wordlet_IPAs = {}

	local split_pattern = "(" .. wordlet_sep .. ")"
	while true do
		local wordlet_sep_at, _, wordlet_sep = find(term, split_pattern, wordlet_start)
		local wordlet_end = wordlet_sep_at and (wordlet_sep_at - 1)
		local part = sub(term, wordlet_start, wordlet_end)

		local stress
		if wordlet_start == 1 then
			stress = "ˈ" -- primary
		elseif wordlet_sep == "/" then
			stress = "ˌ" -- secondary
		else
			stress = hyphenstress
		end

		table.insert(wordlet_IPAs, stress .. fi_IPA_wordlet(part, is_narrow))
		if wordlet_sep_at == nil then
			break
		else
			wordlet_start = wordlet_sep_at + 1
		end
	end

	return wordlet_IPAs
end

-- Generates the broad and optionally also the narrow IPA transcription.
-- If make_narrow is true, the narrow transcription is provided;
-- otherwise it is invalid.
local function fi_IPA_combined_term(term, make_narrow)
	local hyphenstress = "ˌ" -- secondary by default
	local is_prefix, is_suffix

	if find(term, "%/") then
		hyphenstress = tertiary -- tertiary if we have slashes
	end
	
	term = lower(term)
	term, is_suffix = gsub(term, "^%-+", "")
	term, is_prefix = gsub(term, "%-+$", "")

	term = IPA_handle_diphthongs(term)
	
	-- Run IPA for wordlets.
	local IPA = table.concat(fi_IPA_termwordlets(term, false, hyphenstress))
 	
	-- Replace double letters (vowels or consonants) with single letter plus length sign,
	-- and apply post fixes.
	IPA = apply_post_fixes(gsub(IPA, "(%a)%1", "%1" .. long))

	local IPA_narrow
	if make_narrow then
		local term_narrow = add_rhythmic_stress(term)
		IPA_narrow = table.concat(fi_IPA_termwordlets(term_narrow, true, hyphenstress))
		
		-- Handle * in narrow transcription.
		if find(IPA_narrow, "ˣ", nil, true) then
			IPA_narrow = gsub(IPA_narrow, "ˣ(%)?%s*"..stress_p.."?)((.?)" .. ipa_diacritic .. "*)",
				function (post, after, potential_consonant)
					if potential_consonant == "" then
						if find(post, "^%)") then
							return "ʔ" .. post .. after
						else
							return post .. "(ʔ)" .. after
						end
					elseif ipa_consonants:find(potential_consonant) then
						if #post > 0 then
							local amark = ""
							if plosives:find(sub(after, 1, 1)) then
								amark = unreleased
							end
							return after .. amark .. post .. after
						else
							return post .. after .. long
						end
					else
						return post .. "(ʔ)" .. after
					end
				end)
			IPA_narrow = gsub(IPA_narrow, "%(%)", "")
		end

		-- Apply post fixes.
		IPA_narrow = apply_post_fixes_narrow(apply_post_fixes(IPA_narrow))

		if is_prefix > 0 then
			IPA_narrow = IPA_narrow .. "-"
		end
		if is_suffix > 0 then
			IPA_narrow = "-" .. IPA_narrow
		end
	end
	
	if is_prefix > 0 then
		IPA = IPA .. "-"
	end
	if is_suffix > 0 then
		IPA = "-" .. IPA
	end

	return IPA, IPA_narrow
end

local function fi_IPA_term(term, is_narrow)
	local broad, narrow = fi_IPA_combined_term(term, is_narrow)
	return is_narrow and narrow or broad
end

-- exports for tests etc.
export.add_rhythmic_stress = add_rhythmic_stress
export.fi_IPA_term = fi_IPA_term

-- {{fi-IPA}}
function export.fi_IPA_entrypoint(term)
	if type(term) == "table" then
		term = term:getParent().args[1]
	end
	
	local title = mw.title.getCurrentTitle().text
	
	if not term then
		term = title
	elseif term == "*" then
		term = title .. "*"
	end
	
	local no_count = match(term, " ")
	
	IPA_narrow = fi_IPA_term(term, true)
	IPA = fi_IPA_term(term, false)
	return m_IPA.format_IPA_full {
		lang = require("Module:languages").getByCode(langcode),
		items = {{pron = "/" .. IPA .. "/"}, {pron = "[" .. IPA_narrow .. "]"}},
		no_count = no_count,
	}
end

-- <<IPA END>>

local function cleanup_title(x)
	return lower(gsub(x, "–", "-"))
end

-- these two functions are used to build Cartesian products of triplets XAB,
-- such that e.g. (X1, A1, B1), (X2, A2, B2), (X3, _, _)
-- 		creates the combinations
--			X1A1X2A2X3, X1A1X2B2X3, X1B1X2A2X3, X1B1X2B2X3
local function cartesian_make(parts, n)
	local result = parts[1][1]
	local k = 1
	for i = 2, #parts do
		if m_bit32.band(n, k) > 0 then
			result = result .. parts[i - 1][3] .. parts[i][1]
		else
			result = result .. parts[i - 1][2] .. parts[i][1]
		end
		k = m_bit32.lshift(k, 1)
	end
	return result
end

local function cartesian_combine(parts)
	m_bit32 = m_bit32 or require("Module:bit32")
	
	local n = m_bit32.lshift(1, #parts - 1)
	local results = {}
	for i = 0, n - 1 do
		table.insert(results, cartesian_make(parts, i))
	end
	return results
end

local function split_by_optional_break(word, only_breaking_diphthongs, prefer_hyphen_first)
	local parts = {}
	local i = 1
	local found = false

	while true do
		local j, je = find(word, "%([.-]%)", i)
		if j == nil then break end
		local allow_break = true
		if only_breaking_diphthongs then
			local prefix = sub(word, i, j - 1)
			local diphthong = sub(word, j - 1, j - 1) .. sub(word, j + 3, j + 3)
			-- if a dot, only when breaks a diphthong
			allow_break = sub(word, j + 1, j + 1) ~= "." or (
				-- never a diphthong if two vowels precede
				not match(prefix, vowel .. vowel .. "$")
				and vowel_sequences[diphthong])
		end
		if allow_break then
			found = true
			-- prefer dots last if vowels on both sides
			local prev = sub(word, i, j - 1)
			local symbol = sub(word, j + 1, j + 1)
			local prefer_first
			if symbol == "." then
				prefer_first = not (match(prev, vowel .. "$") and match(word, "^" .. vowel, j + 3))
			elseif symbol == "-" then
				prefer_first = prefer_hyphen_first
			end
			if prefer_first then
				table.insert(parts, { prev, symbol, "" })
			else
				table.insert(parts, { prev, "", symbol })
			end
		end
		i = je + 1
	end

	if not found then return { word } end

	table.insert(parts, { sub(word, i), "", "" })

	return cartesian_combine(parts)
end

local function get_autohyphenate_forms(word, title)
	word = gsub(word, "%([*ˣ:ː]%)", "")
	word = gsub(word, "(.)ː", "%1%1")
	word = gsub(word, "[" .. ipa_symb .. "ˣ*]", "")
	word = gsub(word, "/", "-")
	word = gsub(word, "^-", "")
	word = gsub(word, "ŋn", "gn")

	if lower(title) == title then
		word = lower(word)
	else
		-- find letters in title and try to match them
		local letters = {}
		for letter in gmatch(title, "%a") do
			table.insert(letters, letter)
		end

		local respelled = ""
		local letter_index = 1

		for character in gmatch(word, ".") do
			if match(character, "%a") then
				local next_letter = letters[letter_index]
				if lower(next_letter) == lower(character) then
					respelled = respelled .. next_letter
					letter_index = letter_index + 1
				else
					respelled = respelled .. character
				end
			else
				respelled = respelled .. character
			end
		end

		word = respelled
	end

	return split_by_optional_break(word, nil, true)
end

local function get_autorhyme_forms(word)
	word = lower(word)
	word = gsub(word, "%([*ˣ:ː]%)", "")
	-- applies gemination mid-word for rhymes
	word = gsub(word, "[*ˣ](" .. consonant .. ")", "%1ː")
	word = gsub(word, "[*ˣ]", "")
	-- convert long vowel marks
	word = gsub(word, "(.)ː", "%1%1")
	-- remove IPA symbols and normalize +/ to -
	word = gsub(word, "[" .. ipa_symb .. "]", "")
	word = gsub(word, "[/+]", "-")
	return split_by_optional_break(word, nil, true)
end

local function generate_rhyme(word)
	-- convert syllable weight to hyphen for next routine
	-- (just in case these are included manually... even if they shouldn't be)
	local fmtword = gsub(word, "[ˈˌ"..tertiary.."]", "-")
	fmtword = gsub(word, "'", ".")
	
	local sylcount = #syllabify(fmtword, ".")

	-- get final wordlet of a compound word
	local last_hyph = find(fmtword, "%-[^%-]*$") or 0
	local last_wordlet = sub(fmtword, last_hyph + 1)
	-- use fi-IPA rhythmic stress generator
	local stressed = add_rhythmic_stress(last_wordlet)
	-- find last stress mark and extract everything after that
	local res = match(stressed, ".*" .. stress_indicator .. "(.*)")
	if not res or res == "" then res = last_wordlet end
	
	-- remove initial consonants, convert to IPA, remove IPA symbols
	res = gsub(res, "^%.", "")
	res = gsub(res, "^" .. consonant .. "+", "")
	res = fi_IPA_combined_term(res, false)
	res = gsub(res, "[" .. ipa_symb .. "]", "")
	res = gsub(res, "^%.", "")
	
	return res, sylcount
end

-- for testing
export.generate_rhyme = generate_rhyme

local function add_trivowel_dots(pron)
	-- find sequences of >= 3 vowels and record their indices
	local toggles = {}
	local scan = 1
	while true do
		local i0, i1 = find(pron, "[aeiouyäö][aeiouyäö][aeiouyäö]+", scan)
		if i0 == nil then break end
		table.insert(toggles, i0 + 1)
		table.insert(toggles, i1)
		scan = i1 + 1
	end
	table.insert(toggles, len(pron) + 1)

	-- generate hyphenation, and add dots within multivowel sequences
	local dots = {}
	local hyph = syllabify(pron, true)
	local index = 0
	local recons = ""
	scan = 1
	for _, hpart in ipairs(hyph) do
		index = index + len(hpart)
		while index >= toggles[scan] do
			scan = scan + 1
		end
		recons = recons .. hpart
		if scan % 2 == 0 then
			recons = recons .. "."
		end
	end

	return recons
end

local function pron_equal(title, pron)
	if not pron or pron == "" then
		return true
	end
	
	-- handle slashes, pluses and quotes as hyphens
	pron = gsub(pron, '[/+"]', "-")
	-- remove optional lengthening/shortening/syllable break/gemination, should not cause any issues
	pron = gsub(pron, "%([*ˣ.:ː-]%)", "")
	-- remove gemination asterisks and syllable separating dots
	pron = gsub(pron, "[*%." .. syl_override_start .. syl_override_end .. "]", "")
	-- map existing glottal stops to apostrophes
	pron = gsub(pron, "%(?ʔ%)?", apostrophe)
	-- /ŋn/ for /gn/ is fine
	pron = gsub(pron, "ŋn", "gn")
	-- remove hyphens but also apostrophes right after hyphens
	-- (so that glottal stop is allowed after hyphen separating two same vowels)
	pron = gsub(pron, "-" .. apostrophe .. "?", "")
	title = gsub(cleanup_title(title), "-", "")
	
	return pron == lower(title)
end

local function pron_equal_special_cases(title, pronunciation)
	if find(title, "ruoa") then
		-- very common exception - support it
		local pronunciation_ruua = gsub(pronunciation, "ruua", "ruoa")
		if pron_equal(title, lower(pronunciation_ruua)) then
			return pronunciation_ruua
		end
	end
	-- fall through to return nil
end

local bracket_delimiters = {
	["/"] = "[%/]",
	["-"] = "[%/%-]",
	['"'] = '[%/%-%"]',
}

local function treat_brackets(pronunciation_with_brackets)
	local result = {}
	local cursor = 1
	
	while true do
		local bracket_start, bracket_end, bracket_text = find(pronunciation_with_brackets, "(%b[])", cursor)
		if bracket_start == nil then
			table.insert(result, sub(pronunciation_with_brackets, cursor))
			break
		end

		local delimiter = match(bracket_text, '^.([%/"])')
		-- find last slash/hyphen
		local terminate = match(sub(pronunciation_with_brackets, 1, bracket_start), "^.*()" .. bracket_delimiters[delimiter or "-"])
		if cursor > 1 and (not terminate or terminate < cursor) then
			error("bracket {} replacement string had no matching delimiter " .. (delimiter or "-"))
		end
		if terminate then
			table.insert(result, sub(pronunciation_with_brackets, cursor, terminate))
		end

		-- tag replacement with overrides for syllabification
		table.insert(result, syl_override_start .. sub(bracket_text, delimiter and 3 or 2, -2) .. syl_override_end)
		cursor = bracket_end + 1
	end

	return table.concat(result)
end

local function convert_quotes_to(syllabification_cache, text, stress, nostress)
	if find(text, '"') then
		if nostress and stress ~= nostress then
			-- syllabify, and determine which way to go
            -- output nostress if the preceding syllable was stressed,
            -- otherwise output stress

			local syllables = syllabification_cache[text]
			if not syllables then
				syllables = syllabify(text, true)
				syllabification_cache[text] = syllables
			end
			local last_stressed = false
			
			for i, syl in ipairs(syllables) do
				-- first syllable always stressed
				if i == 1 or match(syl, "^" .. stress_indicator) then
					last_stressed = true
				else
					if match(syl, '^"') then
						local next_stressed = i == #syllables or match(syllables[i + 1], "^" .. stress_indicator)
						syllables[i] = gsub(syl, '^"', (last_stressed or next_stressed) and nostress or stress)
					end
					last_stressed = false
				end
			end
			
			text = table.concat(syllables)
		else
			text = gsub(text, '"', stress)	
		end
	end
	return text
end

local function fi_hyphenation_from_syllabification(syl)
	syl = gsub(syl, '[ /"]', "-")
	syl = gsub(syl, "[+'’]", ".")
	
	-- all syllable boundaries are now - or .
	-- first, ban any dots adjacent to a hyphen
	syl = gsub(syl, "%.([^.]-)%-", "%1-")
	syl = gsub(syl, "%-([^.]-)%.", "-%1")
	
	-- then ban any dots that only have a single vowel
	syl = gsub(syl, "^(" .. vowel .. ")%.", "%1")
	syl = gsub(syl, "%.(" .. vowel .. ")$", "%1")
	
	-- in the case x.V.y, technically both xV.y and x.Vy are valid
	-- but here, we only allow xV.y, because there is no way to mark
	-- "allow this or this" without listing all allowed options separately
	-- which can result in exponentially many hyphenations shown
	syl = mw.ustring.gsub(syl, "%.(" .. vowel .. ")%.", "%1.")

	return mw.text.split(syl, "[.-]")
end

function export.show(frame)
	local title = mw.title.getCurrentTitle().text
	local lang = require("Module:languages").getByCode(langcode)

	local pronunciation = { "" }
	local ipa = { nil }
	local audio = { }
	local qualifiers = { }
	local rhymes = { nil }
	local syllabifications = { nil }
	local homophones = { }
	local syllabification_labels = { }
	local rhyme_labels = { }
	local homophone_labels = { }
	local nohyphen = false
	local norhymes = false
	local compound_suffix = false
	local categories = { }
	
	if type(frame) == "table" then
		local params = {
			[1] = { list = true, default = "", allow_holes = true },
			
			["ipa"] = { list = true, default = nil, allow_holes = true },
			["h"] = { list = true, default = nil, allow_holes = true }, ["hyphen"] = {},
			["r"] = { list = true, default = nil, allow_holes = true }, ["rhymes"] = {},
			["a"] = { list = true, default = nil }, ["audio"] = {},
			["ac"] = { list = true, default = nil }, ["caption"] = {},
			["hh"] = { default = "" }, ["homophones"] = {},
			
			["q"] = { list = true, default = nil, allow_holes = true },
			["hp"] = { list = true, default = nil, allow_holes = true },
			["rp"] = { list = true, default = nil, allow_holes = true },
			["hhp"] = { list = true, default = nil, allow_holes = true },
			
			["nohyphen"] = { type = "boolean", default = false },
			["norhymes"] = { type = "boolean", default = false },
			["csuffix"] = { type = "boolean", default = false },
			
			["title"] = {}, -- for debugging or demonstration only
		}
		
		local args, further = require("Module:parameters").process(frame:getParent().args, params, true)
		
		title = args["title"] or title
		pronunciation = args[1]
		ipa = args["ipa"]
		syllabifications = args["h"]
		rhymes = args["r"]
		qualifiers = args["q"]
		syllabification_labels = args["hp"]
		rhyme_labels = args["rp"]
		nohyphen = args["nohyphen"]
		norhymes = args["norhymes"]
		compound_suffix = args["csuffix"]
		homophones = mw.text.split(args["hh"], ",")
		homophone_labels = args["hhp"]
		
		-- hacks; if the first parameter is empty then we want an empty string
		if args[2] and args[1] == nil then args[1] = "" end
		if ipa[2] and ipa[1] == nil then ipa[1] = "" end

        -- clear homophones if empty
		if #homophones == 1 and homophones[1] == "" then homophones = {} end

		if args["hyphen"] then syllabifications[1] = args["hyphen"] end
		if args["rhymes"] then rhymes[1] = args["rhymes"] end
		if args["homophones"] then homophones = mw.text.split(args["homophones"], ",") end
		
		local audios = args["a"]
		local captions = args["ac"]
		if args["audio"] then audios[1] = args["audio"] end
		if args["captions"] then captions[1] = args["caption"] end
		
		for i, audiofile in ipairs(audios) do
			if audiofile then
				table.insert(audio, {lang = lang, file = audiofile, caption = captions[i]})
			end
		end
	end

	for i, p in ipairs(pronunciation) do
		if p == "" or p == "^" then
			pronunciation[i] = cleanup_title(title)
		elseif p == "*" or p == "(*)" then
			pronunciation[i] = cleanup_title(title) .. p
		elseif find(p, "[!#]") then
			p = gsub(p, "t!s", "ts")
			p = gsub(p, "t#s", "ts")
			-- TODO deprecate and remove the two codes below
			p = gsub(p, "%.%!", "(.)")
			p = gsub(p, "%-%!", "-")
			pronunciation[i] = p
		end
	end
	
	-- make sure #pronunciation >= #IPA, because we use #pronunciation
    -- as the source of how many pronunciations we have
	for i, p in ipairs(ipa) do
		if not pronunciation[i] then
			pronunciation[i] = ""
		end
	end
	
    -- whether hyphenation and rhyme data can be generated automatically
	local can_auto_hyphenate = true
    local can_auto_rhyme = true
    -- when generating rhyme or syllabification data automatically, use
    -- model_pronunciation as the source of pronunciation data
	local model_pronunciation

	local i
	-- preprocessing
	local split_optional_breaks = false
	local do_not_split_optional_breaks = false
	local model_pronunciation_index = 1
	local brackets = false
	i = 1
	while i <= #pronunciation do
		-- split (.), but only if there is only one
		local dot_start, dot_end = find(pronunciation[i], "%(%.%)")
		if dot_start then
			split_optional_breaks = true
			if find(pronunciation[i], "%(%.%)", dot_end) then
				do_not_split_optional_breaks = true
			end
		end

		-- add syllable break Vii -> V.ii
		if find(pronunciation[i], "[aeouyäö]ii") then
			pronunciation[i] = gsub(pronunciation[i], "([aeouyäö])(ii)", "%1.%2")
		end

		-- handle % (long consonant usually pronounced as short)
		if find(pronunciation[i], "%", 1, true) then
			local original = pronunciation[i]
			local short = gsub(original, "%%", "")
			local long = gsub(original, "(.)%%", "%1%1")
			pronunciation[i] = short
			-- use long consonant in model_pronunciation
			if i == model_pronunciation_index then
				model_pronunciation_index = i + 1
			end
			i = i + 1
			table.insert(pronunciation, i, long)
			do_not_split_optional_breaks = true
		end

		if find(pronunciation[i], "[", 1, true) then
			brackets = true
		end

        i = i + 1
	end

	local model_pronunciation = pronunciation[model_pronunciation_index]
	if brackets then
		brackets = model_pronunciation
		-- remove brackets from all pronunciation, keeping only in brackets
		-- for later processing

		for i, p in ipairs(pronunciation) do
			pronunciation[i] = gsub(p, "%b[]", "")
		end
		model_pronunciation = pronunciation[model_pronunciation_index]
	end
	
	local model_pronunciation_max_index = model_pronunciation_index
    -- split a single optional break into two pronunciations
    -- model_pronunciation will still have the (.)
    -- syllabification and rhyme logic can handle it
	if #pronunciation == 1 and split_optional_breaks and not do_not_split_optional_breaks then
		local new_pronunciation = {}
		for _, p in ipairs(pronunciation) do
			-- local split_i = split_by_optional_break(p, true)
			local split_i = split_by_optional_break(p)
			for _, np in ipairs(split_i) do
				table.insert(new_pronunciation, np)
			end
		end
		pronunciation = new_pronunciation
		model_pronunciation_max_index = #pronunciation
	end

	local syllabification_cache = {}

	-- generate IPA rows
	local results = {}
	local has_spaces = match(title, " ") or (model_pronunciation and match(model_pronunciation, " "))
	local is_suffix = match(title, "^-")
	local is_affix = not compound_suffix and (match(title, "-$") or is_suffix)
	for i, p in ipairs(pronunciation) do
		local qual = qualifiers[i] or ""
		
		if #qual > 0 then
			qual = " " .. require("Module:qualifier").format_qualifier(qualifiers[i])
		end
		
		if ipa[i] and ipa[i] ~= "" then
			table.insert(results, "* " .. m_IPA.format_IPA_full {
				lang = lang,
				items = {{pron = ipa[i]}},
				no_count = has_spaces,
			} .. qual)
            -- cannot do automatic rhyme/hyph with IPA pronunciations
            if i <= model_pronunciation_max_index then
				can_auto_hyphenate = false
				can_auto_rhyme = false
			end

		else
            -- do basic cleanup to IPA parameters first:

            -- replace + with - for secondary stress, and colon with long
			if find(p, ":") then p = gsub(p, ":", long) end
			if find(p, "%+") then p = gsub(p, "%+", "-") end

			-- add clarifying dots to three-vowel sequences
			if find(p, "[aeiouyäö][aeiouyäö][aeiouyäö]") then
				p = add_trivowel_dots(p)
			end

			-- handle quotes
			if find(p, '"') then
				p = convert_quotes_to(syllabification_cache, p, "-", ".")
			end
			
			local IPA, IPA_narrow = fi_IPA_combined_term(p, true)
			
			-- multi-word stress
			if has_spaces then
				IPA_narrow = gsub(IPA_narrow, " ([^ˈˌ"..tertiary.."])", " ˈ%1")
				IPA = gsub(IPA, " ([^ˈˌ"..tertiary.."])", " ˈ%1")
			end
			
			-- remove initial stress if suffix
			if is_suffix then
				if compound_suffix then
					IPA_narrow = gsub(IPA_narrow, "^(%-?)ˈ", "%1ˌ")
					IPA = gsub(IPA, "^(%-?)ˈ", "%1ˌ")
				else
					IPA_narrow = gsub(IPA_narrow, "^(%-?)ˈ", "%1")
					IPA = gsub(IPA, "^(%-?)ˈ", "%1")
				end
			end
			
			table.insert(results, "* " .. m_IPA.format_IPA_full {
				lang = lang,
				items = {{pron = "/" .. IPA .. "/"}, {pron = "[" .. IPA_narrow .. "]"}},
				no_count = has_spaces,
			} .. qual)
		end
	end
	
    local should_have_automatic_hyphenation = not (has_spaces or is_affix)
    can_auto_rhyme = can_auto_rhyme and should_have_automatic_hyphenation

    -- terms with spaces, or affixes, never have automatic syllabification/rhymes
	local model_pronunciation_hyphenation
    if not should_have_automatic_hyphenation then
        can_auto_hyphenate = false
    elseif can_auto_hyphenate then
        can_auto_hyphenate = pron_equal(title, lower(model_pronunciation))

		if not can_auto_hyphenate then
			model_pronunciation_hyphenation = model_pronunciation

			if brackets then
				-- check for brackets
				model_pronunciation_hyphenation = treat_brackets(brackets)
				can_auto_hyphenate = pron_equal(title, lower(model_pronunciation_hyphenation))
			end

			if not can_auto_hyphenate then
				-- handle special cases
				local special_cases_checked = pron_equal_special_cases(title, model_pronunciation_hyphenation)
				if special_cases_checked then
					can_auto_hyphenate = true
					model_pronunciation_hyphenation = special_cases_checked
				end
			end
		end
    end
	
	local automatic_syllabifications = false
	if can_auto_hyphenate and not syllabifications[1] then
        -- generate automatic syllabifications
		automatic_syllabifications = true
		local quotes_converted = convert_quotes_to(syllabification_cache, model_pronunciation_hyphenation or model_pronunciation, "(-)")
		local forms = get_autohyphenate_forms(quotes_converted, title)
		local seen_syllabifications = {}
		local syllabification_count = 1
		for _, form in ipairs(forms) do
			if syllabifications[syllabification_count] then break end
			local syllabified = table.concat(syllabify(form, sep_symbols, "dots"))
			if not seen_syllabifications[syllabified] then
				syllabifications[syllabification_count] = syllabified
				seen_syllabifications[syllabified] = true
				syllabification_count = syllabification_count + 1
			end
		end
	elseif #syllabifications == 1 and syllabifications[1] == "-" then
		syllabifications = {}
		should_have_automatic_hyphenation = false
	end

	if can_auto_rhyme and not rhymes[1] then
        -- generate automatic rhymes
		local forms = get_autorhyme_forms(convert_quotes_to(syllabification_cache, model_pronunciation, "-", "."))
		for i, form in ipairs(forms) do
			if rhymes[i] then break end
			rhymes[i] = { generate_rhyme(form) }
		end
	elseif #rhymes == 1 and rhymes[1] == "-" then
		rhymes = {}
	end

	for i, a in ipairs(audio) do
		table.insert(results, "* " .. require("Module:audio").format_audio(a))
	end
	
	-- generate rhyme rows from rhyme data
	if not norhymes then
        local syllabification_split_counts = nil

        -- generate rhyme rows
		if #rhymes > 0 then
			-- merge rhymes if they have identical labels
			local last_label = false
			local new_rhymes = {}
			local new_rhyme_labels = {}
			local current_list = {}
			
			for i, r in ipairs(rhymes) do
				local label = rhyme_labels[i] or ""
				if last_label == label then
					table.insert(current_list, r)
				else
					if #current_list > 0 then
						table.insert(new_rhymes, current_list)
					end
					if last_label ~= false then
						table.insert(new_rhyme_labels, last_label)
					end
					current_list = { r }
					last_label = label
				end
			end
			
			table.insert(new_rhymes, current_list)
			table.insert(new_rhyme_labels, last_label)
			rhymes = new_rhymes
			rhyme_labels = new_rhyme_labels
		end
		
        -- add rhymes
		for i, subrhymes in ipairs(rhymes) do
			local label = ""
			if rhyme_labels[i] and #rhyme_labels[i] > 0 then
				label = " " .. require("Module:qualifier").format_qualifier(rhyme_labels[i])
			end

			if #subrhymes >= 1 then
                -- indexed by rhyme first and syllable count second
				local existing_rhymes = {}
				local seen_rhymes = {}
                -- if at least one rhyme doesn't specify the syllable count,
                -- we will try to determine it automatically
				local must_syllabify = false
                local syllable_count
                -- final syllable counts not passed as part of any rhyme
				local fallback_syllable_counts = nil

				for _, rhyme in ipairs(subrhymes) do
					if type(rhyme) == "table" then
						rhyme, syllable_count = unpack(rhyme)
					elseif find(rhyme, "%/") then
						-- parse syllable count
						rhyme, syllable_count = match(rhyme, "(.+)/(.+)")
						syllable_count = tonumber(syllable_count)
					else
						syllable_count = nil
					end
						
					if syllable_count then
						local existing_rhyme = seen_rhymes[rhyme]
						if not existing_rhyme then
                            -- add the rhyme if not present
							local new_rhyme = { rhyme = rhyme, num_syl = {syllable_count} }
							table.insert(existing_rhymes, new_rhyme)
							seen_rhymes[rhyme] = { [syllable_count] = true, object = new_rhyme }
						elseif not existing_rhyme[syllable_count] then
                            -- add the syllable count to an existing rhyme
							table.insert(existing_rhyme.object.num_syl, syllable_count)
							existing_rhyme[syllable_count] = true
						end
					else
						must_syllabify = true
						local existing_rhyme = seen_rhymes[rhyme]
						if not existing_rhyme then
							local new_rhyme = { rhyme = rhyme }
							table.insert(existing_rhymes, new_rhyme)
							seen_rhymes[rhyme] = { object = new_rhyme }
						end
					end
				end

				if must_syllabify and (can_auto_hyphenate or not automatic_syllabifications) then
					fallback_syllable_counts = {}
					local seen_syllable_counts = {}
					
					if not manual_syllable_counts then
                        -- generate syllable counts from provided syllabifications
						manual_syllable_counts = {}
						for i, syl in ipairs(syllabifications) do
							if type(syl) == "string" then
								manual_syllable_counts[i] = #mw.text.split(syl, '[' .. sep_symbols .. ']')
							else
								manual_syllable_counts[i] = #syllabifications[i]
							end
						end
					end
					
					-- get all possible syllable counts from syllabifications
					for i, n in ipairs(manual_syllable_counts) do
						if n > 0 and not seen_syllable_counts[n] then
							table.insert(fallback_syllable_counts, n)
							seen_syllable_counts[n] = true
						end
					end
				end

				table.insert(results, "* " .. require("Module:rhymes").format_rhymes(
					{ lang = lang, rhymes = existing_rhymes, num_syl = fallback_syllable_counts }) .. label)
			end
		end
	end

	-- generate homophone rows from homophone data
	if #homophones > 0 then
		local homophone_rows = {}
		for i, homophone in ipairs(homophones) do
			table.insert(homophone_rows, {
                ["term"] = homophone,
                ["qualifiers"] = homophone_labels[i] and { homophone_labels[i] } or nil
            })
		end
		table.insert(results, "* " .. require("Module:homophones").format_homophones(
			{ lang = lang, homophones = homophone_rows }))
	end

	-- generate hyphenation rows from syllabification data
	if not nohyphen and #syllabifications > 0 then
		local hyphenation_rows = {}
        local seen_hyphenations = {}
		for i, syllabification in ipairs(syllabifications) do
            local hyphenation = fi_hyphenation_from_syllabification(syllabification)
			local hyphenation_key = table.concat(hyphenation, ".")

			-- deduplicate hyphenations
			local hyphenation_row = seen_hyphenations[hyphenation_key]
			if not hyphenation_row then
				hyphenation_row = { 
					["hyph"] = hyphenation,
					["qualifiers"] = nil
				}
				seen_hyphenations[hyphenation_key] = hyphenation_row
				table.insert(hyphenation_rows, hyphenation_row)
			end
			if syllabification_labels[i] then
				hyphenation_row.qualifiers = hyphenation_row.qualifiers or { }
				table.insert(hyphenation_row.qualifiers, syllabification_labels[i])
			end
		end
		table.insert(results, "* " .. require("Module:hyphenation").format_hyphenations(
			{ lang = lang, hyphs = hyphenation_rows, caption = "Hyphenation<sup>([[Appendix:Finnish hyphenation|key]])</sup>" }))
	end
	
	if should_have_automatic_hyphenation and not syllabifications[1] then
		table.insert(categories, "fi-pronunciation missing automatic hyphenation")
	end	
	
	return table.concat(results, "\n") .. require("Module:utilities").format_categories(categories, lang)
end

return export