# morph-ga.txt: Morphology rules for Irish.
# Copyright (C) 2004-2007 Kevin P. Scannell <kscanne@gmail.com>
#
# This is free software; see the file COPYING for copying conditions. There
# is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR
# PURPOSE.
#
# with the exception of some of the non-standard verb endings,
# each rule decreases number of caps or length => recursion will stop.
#
# each substitution is applied globally (though this is only relevant
# for the few rules that aren't anchored at ^ or $
#
# There is no need to encode the regular inflectional morphology of
# Irish here because it is already handled by the C++ code that generates
# the basic lexicon. So if "imirceacha" is not in the lexicon (it is)
# "imirceach" won't be either. On the other hand it can be quite useful
# to include *derivational* morphology; so if "gaimbíneachas" is not
# in the lexicon (it is) there is some hope that by stripping the
# "eachas" it will find the root "gaimbín".
^([ACEFH-LO-SU-ZÁÉÍÓÚ])([A-ZÁÉÍÓÚ'-]*[A-ZÁÉÍÓÚ][A-ZÁÉÍÓÚ'-]*)$ $1\L$2 -1 <.+> <&> # ÉIRE; potentially eclipsing init chars require more care:
^B([A-GI-OQ-ZÁÉÍÓÚ][A-ZÁÉÍÓÚ'-]*)$ B\L$1 -1 <.+> <&>
^B'([AEIOUÁÉÍÓÚF][A-ZÁÉÍÓÚ'-]*)$ B'\L$1 -1 <.+> <&>
^BH([A-EG-ZÁÉÍÓÚ][A-ZÁÉÍÓÚ'-]*)$ Bh\L$1 -1 <.+> <&>
^D([A-SU-ZÁÉÍÓÚ][A-ZÁÉÍÓÚ'-]*)$ D\L$1 -1 <.+> <&>
^D'([AEIOUÁÉÍÓÚF][A-ZÁÉÍÓÚ'-]*)$ D'\L$1 -1 <.+> <&>
^G([ABD-ZÁÉÍÓÚ][A-ZÁÉÍÓÚ'-]*)$ G\L$1 -1 <.+> <&>
^M([AC-ZÁÉÍÓÚ][A-ZÁÉÍÓÚ'-]*)$ M\L$1 -1 <.+> <&>
^N([ABCEFH-ZÁÉÍÓÚ-][A-ZÁÉÍÓÚ'-]*)$ N\L$1 -1 <.+> <&>
^T([A-RT-ZÁÉÍÓÚ-][A-ZÁÉÍÓÚ'-]*)$ T\L$1 -1 <.+> <&>
^([ACEFH-LO-SU-ZÁÉÍÓÚ]) \l$1 -1 <.+> <&> # Príomhach
^B([^pPh]) b$1 -1 <.+> <&>
^Bh([^fF]) bh$1 -1 <.+> <&>
^D([^tT]) d$1 -1 <.+> <&>
^G([^cC]) g$1 -1 <.+> <&>
^M([^bB]) m$1 -1 <.+> <&>
^N([^dDgG]) n$1 -1 <.+> <&>
^T([^sS]) t$1 -1 <.+> <&>
(...)-([A-ZÁÉÍÓÚ]) $1-\l$2 -1 <.+> <&> # Príomh-Aire
^b'([AEIOUÁÉÍÓÚF]) b'\l$1 -1 <.+> <&>
^bP([A-ZÁÉÍÓÚ'-]+)$ bp\L$1 -1 <.+> <&> # bPRÍOMH-AIRE
^bP bp -1 <.+> <&> # bPríomhach
^bhF([A-ZÁÉÍÓÚ'-]+)$ bhf\L$1 -1 <.+> <&>
^bhF bhf -1 <.+> <&>
^d'([AEIOUÁÉÍÓÚF]) d'\l$1 -1 <.+> <&>
^dT([A-ZÁÉÍÓÚ'-]+)$ dt\L$1 -1 <.+> <&>
^dT dt -1 <.+> <&>
^gC([A-ZÁÉÍÓÚ'-]+)$ gc\L$1 -1 <.+> <&>
^gC gc -1 <.+> <&>
^h([AEIOUÁÉÍÓÚ])([A-ZÁÉÍÓÚ'-]+)$ h$1\L$2 -1 <.+> <&> # hÉIREANN
^h([AEIOUÁÉÍÓÚ]) h\l$1 -1 <.+> <&>
^m'([AEIOUÁÉÍÓÚF]) m'\l$1 -1 <.+> <&>
^mB([A-ZÁÉÍÓÚ'-]+)$ mb\L$1 -1 <.+> <&>
^mB mb -1 <.+> <&>
^n([AEIOUÁÉÍÓÚ])([A-ZÁÉÍÓÚ'-]+)$ n$1\L$2 -1 <.+> <&>
^n([AEIOUÁÉÍÓÚ]) n-\l$1 -1 <.+> <&>
^nD([A-ZÁÉÍÓÚ'-]+)$ nd\L$1 -1 <.+> <&>
^nD nd -1 <.+> <&>
^nG([A-ZÁÉÍÓÚ'-]+)$ ng\L$1 -1 <.+> <&>
^nG ng -1 <.+> <&>
^tS([A-ZÁÉÍÓÚ'-]+)$ ts\L$1 -1 <.+> <&>
^tS ts -1 <.+> <&>
^t([AEIOUÁÉÍÓÚ])([A-ZÁÉÍÓÚ'-]+)$ t$1\L$2 -1 <N.*> <&> # tUASAL
^t([AEIOUÁÉÍÓÚ]) t-\l$1 -1 <N.*> <&>
# these aren't in aspell db by default; better to do early (mb'amhlaidh)
^mb' b' -1 <.+> <&>
^d'f([aeiouáéíóú]) d'fh$1 2 <.+> <&>
# do these early - avoids mistakes with "aighthe" (==aghaidheanna)
# and note that first rule must precede second to get "-aithe" pp's correct
# IMPORTANT - used for -ughadh endings e.g., -> -aghadh which goes to ú below
u(i?[dg]h) a$1 1 <.+> <&> # see next few
i[dg]h(th?.) i$1 1 <.+> <&> # marcaidhthe, masluighthe, beannuighth?ear
#############################################################################
################## CAIGHDEÁN OIFIGIÚIL SPELLING REFORMS #####################
#############################################################################
# modernize prefixes (no need to trap mutated versions, stripped below
# full list from OD77 is in gaeilge/diolaim/x/OD77alt-prefixes
^h?aith ath 1 <.+> <&>
^h?aird([^e]) ard$1 1 <.+> <&>
# handled by more general "nn" rule below
#^h?anns ans 1 <.+> <&>
^dh' d' 1 <.+> <&>
^h- h 1 <.+> <&> # CO p.126
^n-([AEIOUÁÉÍÓÚ]) n$1 1 <.+> <&> # CO p.125
^t-([AEIOUÁÉÍÓÚ]) t$1 1 <.+> <&> # CO p.125
^h?aoin aon 1 <.+> <&>
^h?ana- an- 1 <.+> <&> # ana-eagraithe
#^[Áá]rd(-?..) ard$1 1 <.+> <&>
^h?ath-?th at 1 <[ANV].*> <&>
^bh' b' 1 <.+> <&> # bh'fhiú -> b'fhiú
^bain ban 1 <.+> <&>
^báin bán 1 <.+> <&>
^bairr barr 1 <.+> <&>
^baoith baoth 1 <.+> <&>
^beig beag 1 <.+> <&>
^béil béal 1 <.+> <&>
^bioth bith 1 <.+> <&>
^boig bog 1 <.+> <&>
^boinn bonn 1 <.+> <&>
^boirr borr 1 <.+> <&> # boirrphéist...
# handled by more general ll rule below
#^bolls bols 1 <.+> <&>
^buadh bua 1 <.+> <&>
^buain buan 1 <.+> <&> # buainseasamh...
^caim cam 1 <.+> <&> # caimbheart...
# handled by more general tht rule below
#^caitht cait 1 <.+> <&>
^caoimh caomh 1 <.+> <&>
# N.B. makes string longer
#^ceathar ceathair 1 <.+> <&>
^céid céad 1 <.+> <&>
# N.B. makes string longer
#^cinn ceann 1 <.+> <&>
^ceinn ceann 1 <.+> <&> # Di04 ceinn-litir, srl.
^cionn ceann 1 <.+> <&>
^cláir clár 1 <.+> <&> # cláirfhiacail e.g.
^claoidh cloí 1 <.+> <&>
^claoin claon 1 <.+> <&>
^coilg colg 1 <.+> <&> # coilgsheasamh e.g.
^c[oó]imh-mh cóimh 1 <[ANV].*> <&>
^coimh comh 1 <.+> <&> # coimhlíonadh
^cóimhmh cóimh 1 <.+> <&> # cóimhmheas, srl in Di04
^cóimh([^eéií]) comh$1 1 <.+> <&> # cóimh-chealg (Di04)
# general ómh[^a] -> omh$1 rule below
^cómha comha 1 <[ANV].*> <&>
^cómhmh cómh 1 <.+> <&> # cómhmhaith, srl in Di04
^comh-(mh[aáoóuú]) có$1 1 <[ANV].*> <&>
^comh-(mh[eéií]) cói$1 1 <[ANV].*> <&>
^coimh-n cóin 1 <[ANV].*> <&>
^c[oó]mh-?(n[aáoóuú]) có$1 1 <[ANV].*> <&>
^c[oó]mh-?(n[eéií]) cói$1 1 <[ANV].*> <&>
^có-(..) comh$1 1 <[ANV].*> <&>
^có([bcdfgpt]h) comh$1 1 <[ANV].*> <&>
^c[ou]ir([pr]) cor$1 1 <.+> <&>
^crainn crann 1 <.+> <&>
^crioth crith 1 <.+> <&>
^croim crom 1 <.+> <&> # croimleac e.g.
^cruai?dh crua 1 <.+> <&>
^cúil cúl 1 <.+> <&>
^dá- dé 1 <.+> <&> # common in Di04 e.g. dá-bheathach
^daill dall 1 <.+> <&> # daillintinn
^daoir([^s]) daor 1 <.+> <&>
# first spelling of dea- is from na Grianna
^déagh- dea- 1 <.+> <&>
^de[áai]gh- dea- 1 <.+> <&>
^de[áai]gh([^-]) dea-$1 1 <.+> <&>
^deir([bg]) dear$1 1 <.+> <&> # deirbhnialas, deirgmheisce
^dío- dí- 1 <.+> <&> # Di04 common. dío-chuimhne
^diubhr diúr 1 <.+> <&>
^dligh(i?) dlí 1 <.+> <&>
^doi- do- 1 <.+> <&> # doi-bhriathar, etc. Di04
^doinn donn 1 <.+> <&>
^droich droch 1 <.+> <&>
^druim droim 1 <.+> <&>
^duibh dubh 1 <.+> <&>
^dubh- dú 1 <.+> <&>
# handled by more general prefix+gc rule below
#^éagc éag 1 <.+> <&>
#^éigc éig 1 <.+> <&>
# handled by more general prefix+dt rule below
#^éadt éad 1 <.+> <&>
#^éidt éid 1 <.+> <&>
# handled by more general ll rule below
#^falls fals 1 <.+> <&>
^fágbh fág 1 <.+> <&>
^h?eadar idir 1 <.+> <&>
^h?eidir idir 1 <.+> <&>
^fiadh fia 1 <.+> <&>
^fír- fíor 1 <.+> <&> # include hyphen so len is same
^flith frith 1 <.+> <&>
^frioth([^á]) frith$1 1 <.+> <&>
^fó- fo- 1 <.+> <&>
^foir- for- 1 <.+> <&> # foir-chéim
^fuair- fuar- 1 <.+> <&> # fuair-chrith
# N.B. makes string longer
#^gamh geamh 1 <.+> <&>
# N.B. makes string longer
^girr gearr 1 <.+> <&>
^gairbh garbh 1 <.+> <&>
^géir géar 1 <.+> <&>
^gnáith gnáth 1 <.+> <&>
^gobh gabh 1 <.+> <&>
^héala([^aeiouáéíóú]+[eéií]) héili$1 1 <.+> <&> # OD77
^héala([^aeiouáéíóú]+[aáoóuú]) héilea$1 1 <.+> <&>
^h?iair iar 1 <.+> <&>
^h?iarann iarn 1 <.+> <&>
^h?iodar idir 1 <.+> <&> # iodarchaidreamh
^h?iol([^ar]) il$1 1 <.+> <&>
^h?iomshlá iomlá 1 <.+> <&> # exception to next
^h?iom([^aálpr]) im$1 1 <.+> <&>
^h?ion in 1 <.+> <&>
# handled by more general nn rule below
#^h?ionn([rs]) ion$1 1 <.+> <&>
^([ls])áimh $1ámh 1 <.+> <&>
^láin lán 1 <.+> <&> # láinchinnte
# handled by more general sb rule below
#^léasb léasp 1 <.+> <&>
^leig([^h]) lig$1 1 <.+> <&> # leigint, but not leigheas words
# risky?
^leith- leath- 1 <.+> <&>
^loim lom 1 <.+> <&> # loimeasna
^lóir leor 1 <.+> <&>
^luaith([^r]) luath$1 1 <.+> <&> # luaithintinn
# handled by more general sg rule below
#^luasg luasc 1 <.+> <&>
# N.B. makes string longer
^lubh luibh 1 <.+> <&> # lubhghort
^lui([mn]) loi$1 1 <.+> <&> # luinnir->loinnir
^lúith lúth 1 <.+> <&>
^maill mall 1 <.+> <&>
^maoil maol 1 <.+> <&>
^maoith([^n]) maoth$1 1 <.+> <&> # maoithintinneach
^mairbh marbh 1 <.+> <&> # mairbhghin
# only two math- words anyway?
#^magh math 1 <.+> <&>
^meadhón meán 1 <.+> <&>
^mean meán 1 <.+> <&>
^mh'([aeiouáéíóú].) m'$1 1 <.+> <&>
# N.B. makes string longer
^min- mion- 1 <.+> <&>
^mío- mí- 1 <.+> <&>
^mío([bcdfgmpst]h) mí$1 1 <[ANV].*> <&> # míochothrom
^(mh?í)-([^aeiouáéíóú].) $1$2 1 <[ANV].*> <&>
# no midh- words anyway?
#^miodh midh 1 <.+> <&>
^móir mór 1 <.+> <&>
^naoimh naomh 1 <.+> <&>
^neamh-([^m].) neamh$1 1 <[ANV].*> <&> # try stripping hyphen before stripping whole thing!
^neimh(..) neamh$1 1 <.+> <&> # Di04
^nea-mb neamhbh 1 <.+> <&>
^nea- neamh- 1 <.+> <&> # strip hyphen later
^nea([cdfgpt]h) neamh$1 1 <.+> <&>
^h?ói([gr]) ó$1 1 <.+> <&>
# handled by more general nn rule below
#^pannc panc 1 <.+> <&>
#^poinnt point 1 <.+> <&>
^pont punt 1 <.+> <&>
^prímh príomh 1 <.+> <&>
^réidh ré 1 <.+> <&>
^réimh réamh 1 <.+> <&>
^r[íi]o?gh([^n]) rí$1 1 <.+> <&> # righdhamhna, ríogh-choróin, but not "righnigh", "righne", etc.
^ro- ró- 1 <.+> <&>
^rói- ró- 1 <.+> <&>
^roighn righn 1 <.+> <&>
^roimh réamh 1 <.+> <&>
^ruadh rua 1 <.+> <&>
^rúin rún 1 <.+> <&>
^sáir sár 1 <.+> <&> # sáireolas
^saoghl saol 1 <.+> <&>
^saoibh saobh 1 <.+> <&>
^saoir- saor- 1 <.+> <&>
^sclug sclog 1 <.+> <&>
^scoith scoth 1 <.+> <&> # scoithdhearg
^seana- sean- 1 <.+> <&> # strip hyphen later
^sein- sean- 1 <.+> <&> # strip hyphen later
^siod sead 1 <.+> <&>
^síodh sí 1 <.+> <&>
# N.B. makes it longer
^sír- síor- 1 <.+> <&>
^smúid smúit 1 <.+> <&>
^soi- so- 1 <.+> <&> # soi-bhriste, etc. Di04
# handled by more general ll rule below
#^soills soils 1 <.+> <&>
# handled by more general nn rule below
#^sonnr sonr 1 <.+> <&>
^sr[áa]ic srac 1 <.+> <&>
# N.B. makes string longer
^srang sreang 1 <.+> <&>
^sream sram 1 <.+> <&>
# N.B. makes string longer - risky, plus no "sruim-" words at all?
#^srim sruim 1 <.+> <&>
^stáit stát 1 <.+> <&> # stáitseirbhís
^taidhbh taibh 1 <.+> <&>
# handled by more general sb rule below
#^taisb taisp 1 <.+> <&>
^teasb([^h]) taispe$1 1 <.+> <&>
^teint tint 1 <.+> <&>
^tele([^aeiouáéíóú]+[eéií]) teili$1 1 <.+> <&> # OD77
^tele([^aeiouáéíóú]+[aáoóuú]) teilea$1 1 <.+> <&>
^tiom tim 1 <.+> <&>
^tiugh tiú 1 <.+> <&>
# risky
#^tiúr tabhar 1 <.+> <&>
# handled by more general rule below
#^treabhth treaf 1 <.+> <&>
^treas tras 1 <.+> <&>
^tréin tréan 1 <.+> <&>
^troim trom 1 <.+> <&>
^tuaith tuath 1 <.+> <&> # tuaithcheantar
^h?uaith uath 1 <.+> <&>
^h?úir úr 1 <.+> <&>
# End of Prefixes - Now general spelling reforms (but not from CO handbook)
# Mostly based on corpus work...
# deán for déan
^(n?dh?)eán $1éan 1 <V.*> <&>
eoracht(a?)$ eoireacht$1 1 <[FN].*> <&> # spaisteoracht
óracht(a?)$ óireacht$1 1 <[FN].*> <&> # fáidheadóracht
éair$ éir 1 <[FN].*> <&> # piléair, paipéair
eor$ eoir 1 <.+> <&>
# 08-11-05, generalize nouns to everything (fraochta e.g.)
chd(a?)$ cht$1 1 <.+> <&> # masc too (achd, m.sh.)
ínteacht(a?)$ íneacht$1 1 <[FN].*> <&> # deisimínteacht, etc.
óanna$ ónna 1 <[FN].*> <&> # common! clóanna, seóanna, gnóanna, srl.
# corpus - Dómhnach, etc., but not rómhilis, etc.
# Handles all ^cómh- prefixes too except ^cómha which we do above explicitly
([^Rr])ómh([^a]) $1omh$2 1 <.+> <&>
^(.)ógh $1ogh 1 <.+> <&> # fóghmhar->foghmhar->fómhar
eamhn iún 1 <.+> <&> # maitheamhnas, breitheamhnas
([^e])amhn $1ún 1 <.+> <&> # rathamhnas
# warning - makes longer
^b'f([aeiouáéíóú]) b'fh$1 1 <.+> <&>
# Now various general spelling reforms from CO handbook
ai(cht?ín) a$1 1 <.+> <&> # CO p.120, smaichtín
sg sc 1 <.+> <&> # CO p.115
sb sp 1 <.+> <&> # CO p.115
sd st 1 <.+> <&> # CO p.115
dn nn 1 <.+> <&> # CO p.114 céadna
é(ir[ld]) e$1 1 <.+> <&> # CO p.123
á(i?r[dlnr]) a$1 1 <.+> <&> # CO p.123-124
ó(i?r[dlnr]) o$1 1 <.+> <&> # CO p.123-124
ú(i?r[dlnr]) u$1 1 <.+> <&> # CO p.123-124
eó eo 1 <.+> <&> # go leór, aibhleóg, etc.
eamh[au]il$ iúil 1 <[ANF].*> <&> # dlightheamhail
eamhla iúla 1 <[ANF].*> <&>
([^e])amh[au]il$ $1úil 1 <[ANF].*> <&> # CO p.102, a2+barúil, etc.
([^e])amhla $1úla 1 <[ANF].*> <&> # "-amhlachta?" too
adht([aá]) at$1 1 <.+> <&> # iadhta, cruadhtán, iarradhtas, srl
aíle$ aíola 1 <.+> <&> # *3ú* decl.
eachaibh$ igh 1 <N pl="y".*> <&> # taoiseachaibh
([^e])achaibh$ $1aigh 1 <N pl="y".*> <&> # Albanachaibh
anaibh$ anna 1 <N pl="y".*> <&> # sgoileanaibh, ceisteanaibh
ánaibh$ áin 1 <N pl="y".*> <&> # uachtaránaibh
aibh$ a 1 <N pl="y".*> <&> # sinagógaibh, fiachaibh
(..[^a])ibh$ $1í 1 <N pl="y".*> <&> # múinteoiribh, etc.
(.)eu $1éa 1 <.+> <&> # sgeul, ceudna, srl.
tch t 1 <.+> <&> # CO p.103
i[dg]hea ío 1 <.+> <&> # CO p.105,107
i[dg]he í 1 <.+> <&> # CO p.104,106
uí([mto]) aí$1 1 <.+> <&> # cialluíonn, ionaduíochta, leasuítear, mionnuím
uith aith 1 <.+> <&> # rialuitheoir
adha ú 1 <.+> <&> # CO p.105
ádh á 1 <.+> <&> # ádhbhar, rádh, etc.
[mb]hth f 1 <.+> <&> # CO p.106
th[mb]h f 1 <.+> <&> # CO p.106
ghai aí 1 <.+> <&> # CO p.107
thch ch 1 <.+> <&> # CO p.108
tht t 1 <.+> <&> # CO p.108
ll([rst]) l$1 1 <.+> <&> # CO p.112-113
nn([cdlrst]) n$1 1 <.+> <&> # CO p.114
ghail$ aíl 1 <N pl="n" gnt="n" gnd="f".*> <&>
ghaile$ aíola 1 <N pl="n" gnt="y" gnd="f".*> <&>
ighil$ íl 1 <N pl="n" gnt="n" gnd="f".*> <&>
ighile$ íola 1 <N pl="n" gnt="y" gnd="f".*> <&>
(ain|[eé][ai]|iar)gc $1g 1 <.+> <&> # CO p.109
e[au]gc éag 1 <.+> <&> # CO p.109
(é[ai])dt $1d 1 <.+> <&> # CO p.109
(dh?í?)omb $1om 1 <.+> <&> # CO p.109
(ai?n)bhf $1bh 1 <.+> <&> # CO p.109 ainbhfiosach
(..)u(i?s)$ $1a$2 1 <.+> <&> # solus, árus, etc.
# some dialect/pre-standard noun/adjective inflections
([oó]ga)í$ $1 1 <N pl="y".*> <&> # duilleogaí
nnaí$ nna 1 <N pl="y".*> <&> # ceisteannaí
(..)dha$ $1tha 1 <.+> <&> # measardha, muinteardha
(..)dha$ $1a 1 <.+> <&> # stuamdha, príomhdha
eói?g$ eog 1 <[NF].*> <&> # nb2
eoig$ eog 1 <.+> <&> # nb2, leave as <.+> for <F>'s
óig$ óg 1 <[NF].*> <&> # nb2
lainn$ lann 1 <[NF].*> <&> # nb2
aisí$ ais 1 <[NF].*> <&> # cuntaisí, bronntanaisí
dth([ea])$ dt$1 1 <.+> <&> # goidthe
# Fórsaíbh -> Fórsaí (dpl)
íbh$ í 1 <[NF].*> <&>
íocha$ í 1 <N pl="y".*> <&> # margaidheacha->margaíocha->margaí
# but note in previous that sometimes this gets interrupted by correct
# plural adjective: cogaidheacha->cogaíocha, or sometimes even correct
# plural noun: claidheacha->claíocha (so want to keep -íocha intermediate)
################## NON/PRE-STANDARD VERB ENDINGS #########################
# future/conditional 2nd declension with -cha?- see CO p.103
e[oó]cha(i?dh)$ eo$1 1 <V.*> <&>
([^e])ócha(i?dh)$ $1ó$2 1 <V.*> <&>
e[oó]chaimi([ds])$ eoimi$1 1 <V.*> <&>
([^e])óchaimi([ds])$ $1óimi$2 1 <V.*> <&>
e[oó]chamuid$ eoimid 1 <V p="y" t="fáist"> <&>
([^e])óchamuid$ $1óimid 1 <V p="y" t="fáist"> <&>
e[oó]chai(nn|dís)$ eoi$1 1 <V p="y" t="coinn"> <&>
([^e])óchai(nn|dís)$ $1ói$2 1 <V p="y" t="coinn"> <&>
e[oó]chthá$ eofá 1 <V p="y" t="coinn"> <&>
([^e])óchthá$ $1ófá 1 <V p="y" t="coinn"> <&>
e[oó]ch(th)?aidhe$ eofaí 1 <V p="n" t="coinn"> <&>
([^e])óch(th)?aidhe$ $1ófaí 1 <V p="n" t="coinn"> <&>
(ó|eo)char$ $1far 1 <V p="n" t="fáist"> <&>
(ó|eo)chthar$ $1far 1 <V p="n" t="fáist"> <&>
faidhear$ far 1 <V p="n" t="fáist"> <&>
# dá n-amharcthá, dá dtugthá, srl
th(eá)$ t$1 1 <V p="y" t="gnáth"> <&>
# 2nd declension Dinneen-style endings
(..)aíd$ $1aíonn 1 <V p="y" t="láith"> <&> # cialluíd
aghadh ú 1 <[NF].*> <&> # broad and slender
iaghadh íodh 1 <[VF].*> <&> # innsiughadh->insíodh
(..)thear$ $1tear 1 <[VF].*> <&> # aut. present
(..)t(te?ar)$ $1$2 1 <[VF].*> <&> # aut. present
ighdar íodar 1 <[VF].*> <&> # broad and slender
ighim ím 1 <[VF].*> <&> # broad and slender
ighidh í 1 <.+> <&> # subjunctive?
ighinn ínn 1 <.+> <&> # imperfect
uigh'e$ aithe 1 <.+> <&> # U019.txt
fiomuid$ fimid 1 <[VF].*> <&>
# Future forms: see OS88 p.215 for these (fe?as, etc. on p.63), also CB p.145
# socróid, dílseoid, 3rd plural forms in achtanna
fa[ds]$ faidh 1 <V p="y" t="fáist"> <&>
fea[ds]$ fidh 1 <V p="y" t="fáist"> <&>
óchas$ óidh 1 <V p="y" t="fáist"> <&> # meabhróchas->meabhrós->...
ói?[ds]$ óidh 1 <V p="y" t="fáist"> <&>
eoi?[ds]$ eoidh 1 <V p="y" t="fáist"> <&>
# Common in achtanna to have -fidh,-fid ending on broad stem, lots of
# other slender/broad mismatches:
([aáoóuú][^aeiouáéíóú]*)fí$ $1faí 1 <V p="n" t="coinn"> <&>
([aáoóuú][^aeiouáéíóú]*)fidh?$ $1faidh 1 <V p="y" t="fáist"> <&>
([eéií][^aeiouáéíóú]*)fadh$ $1feadh 1 <V p="y" t="coinn"> <&>
([eéií][^aeiouáéíóú]*)far$ $1fear 1 <V p="n" t="fáist"> <&>
([eéií][^aeiouáéíóú]*)tar$ $1tear 1 <V p="n" t="láith"> <&> # * not + for chítar
# Future 2nd p., "r" forms noted on OS88 p.216;
# see achtanna for íocfaid/íocfa, bainfid/bainfe
fa$ faidh 1 <V p="y" t="fáist"> <&>
fai[rsd]$ faidh 1 <V p="y" t="fáist"> <&>
fe$ fidh 1 <V p="y" t="fáist"> <&>
fi[rsd]$ fidh 1 <V p="y" t="fáist"> <&>
óis$ óidh 1 <V p="y" t="fáist"> <&>
eois$ eoidh 1 <V p="y" t="fáist"> <&>
# Future/Cond autonomous; unwanted prefix h is common in achtanna
^h([aeiouáéíóú].+fe?ar)$ $1 1 <V p="n" t="fáist"> <&>
^h([aeiouáéíóú].+fa?í)$ $1 1 <V p="n" t="coinn"> <&>
# Past; also see Di27 (present). "r" forms on p.216 (also CB p.145 present)
(...)íos$ $1igh 1 <V p="y" t="caite"> <&>
(...)ai[rs]$ $1 1 <V p="y" t="caite"> <&>
(..[^aeiouáéíóú])i[rs]$ $1 1 <V p="y" t="caite"> <&>
(...)ís$ $1igh 1 <V p="y" t="caite"> <&>
# Thugtaoi in MU44, e.g.
taoi$ taí 1 <V p="n" t="gnáth"> <&>
# present
ains$ ann 1 <V p="y" t="láith"> <&>
ins$ eann 1 <V p="y" t="láith"> <&>
íns$ íonn 1 <V p="y" t="láith"> <&>
# OS88 p.126 present
anns$ ann 1 <V p="y" t="láith"> <&>
íonns$ íonn 1 <V p="y" t="láith"> <&>
# corpus; now handled above
# th(eá|ear)$ t$1 1 <V.*> <&>
famuid faimid 1 <V p="y" t="fáist"> <&>
# -idh, -aidh on 1st declension verbs: Pres. Subj. or 2nd pl. imper (-igí)
# according to Dinneen. Some examples seem to be pres. in corpus,
# e.g. suidhidh, some apparently past e.g. uneclipsed "tigidh"
# Here we go with subjunctive which is the most common by far
(.[^a])idh$ $1e 1 <V p="y" t="foshuit"> <&>
# (..)aidh$ $1a 1 <V p="y" t="foshuit"> <&>
# pre-standard texts commonly have "déanfaimíd", "tugaimís", "rachaimíd"...
(...)imí([ds])$ $1imi$2 1 <V.*> <&>
igidh$ igí 1 <V p="y" t="ord"> <&>
mist$ mis 1 <V.*> <&>
# various -áil endings should be broad, CO p.120
áilim$ álaim 1 <V.*> <&>
áileann$ álann 1 <V.*> <&>
áilfidh$ álfaidh 1 <V.*> <&>
áilfeá$ álfá 1 <V.*> <&>
áilfear$ álfar 1 <V.*> <&>
áilfí$ álfaí 1 <V.*> <&>
ái(lf?)eadh$ á$1adh 1 <V.*> <&>
ái(lf?)im(i[sd])$ á$1aim$2 1 <V.*> <&>
áiligí$ álaigí 1 <V.*> <&>
áilea([dm]ar)$ ála$1 1 <V.*> <&>
ái(lf?)idís$ á$1aidís 1 <V.*> <&>
áltar$ áiltear 1 <V.*> <&>
álta$ áilte 1 <A.*> <&>
álaithe$ áilte 1 <A.*> <&>
álanna$ álacha 1 <N.*> <&> # nb3 pl
#############################################################################
# start some more radical changes, e.g. stripping prefixes completely
#############################################################################
# hyphened prefixes, an-, dea-
^h?an-([bcfgmp]h) $1 0 <[AN].*> <&>
^h?an-([bcfgmp][^h]) $1 2 <[AN].*> <&>
^h?an-([^bcfgmp][^h]) $1 0 <[AN].*> <&>
^h?an-([^bcfgmp]h) $1 2 <[AN].*> <&> # an-dhuine
^dea-([bcdfgmpt]h) $1 0 <[AN].*> <&>
^dea-([bcdfgmpt][^h]) $1 2 <[AN].*> <&>
^dea-(sh[aeiouáéíóúlnr]) $1 0 <[AN].*> <&>
^dea-(s[aeiouáéíóúlnr]) $1 2 <[AN].*> <&>
^dea-(s[^aeiouáéíóúlnrh]) $1 0 <[AN].*> <&> # dea-scéal
^dea-([^bcdfgmpst]) $1 0 <[AN].*> <&>
# other prefixes; most are not independent words (ath, comh, etc.)
# but several are -- these are worth including for several reasons:
# (1) efficiency; morph. analysis happens well before the check for
# compounds/run-togethers (2) allows for a more refined check for
# lenition of the prefixed word or other changes (comh-mh, comh->có, etc.)
^h?ain([deéiílnrst]) $1 0 <[AN].*> <&>
^h?ain([bcfgmp]h) $1 0 <[AN].*> <&>
^h?ath([^bcdfgmpst-]) $1 0 <[ANV].*> <&>
^h?ath([bcdfgmp]h) $1 0 <[ANV].*> <&>
^h?ath(sh[aeiouáéíóúlnr]) $1 0 <[ANV].*> <&>
^h?ath(s[^aeiouáéíóúlnrh]) $1 0 <[ANV].*> <&>
^h?ath-?th th 1 <[ANV].*> <&> # -> "at-" above
^comh([^bcdfgmnpst-]) $1 0 <[ANV].*> <&>
^comh([bcdfgpt]h) $1 0 <[ANV].*> <&>
^comh(sh[aeiouáéíóúlnr]) $1 0 <[ANV].*> <&>
^comh(s[^aeiouáéíóúlnrh]) $1 0 <[ANV].*> <&>
# cómhargadh, cómhalartach; but pre-standard "cómhoibriughadh" too
^có(mh[aáoóuú]) $1 0 <[ANV].*> <&>
^cói(mh[eéií]) $1 0 <[ANV].*> <&>
^có(n[aáoóuú]) $1 0 <[ANV].*> <&>
^cói(n[eéií]) $1 0 <[ANV].*> <&>
^do-([aeiouáéíóú]) $1 0 <[AN].*> <&>
# ^do([lnr]) BELOW
^do([bcdfgmpt]h) $1 0 <[AN].*> <&>
^do(sh[aeiouáéíóúlnr]) $1 0 <[AN].*> <&>
^do(s[^aeiouáéíóúlnrh]) $1 0 <[AN].*> <&>
^do-([aeiouáéíóú]) $1 0 <[ANV].*> <&>
^droch([^bcdfgmpst-]) $1 0 <[AN].*> <&>
^droch-(ch) $1 0 <[AN].*> <&>
^droch([bdfgmpt]h) $1 0 <[AN].*> <&>
^droch(sh[aeiouáéíóúlnr]) $1 0 <[AN].*> <&>
^droch(s[^aeiouáéíóúlnrh]) $1 0 <[AN].*> <&>
^fíor([^bcdfgmprst-]) $1 0 <[ANV].*> <&>
^fíor-(r) $1 0 <[ANV].*> <&>
^fíor([bcdfgmpt]h) $1 0 <[ANV].*> <&>
^fíor(sh[aeiouáéíóúlnr]) $1 0 <[ANV].*> <&>
^fíor(s[^aeiouáéíóúlnrh]) $1 0 <[ANV].*> <&>
^for([^bcdfgmprst-]) $1 0 <[ANV].*> <&>
^for-(r) $1 0 <[ANV].*> <&>
^for([bcdfgmpt]h) $1 0 <[ANV].*> <&>
^for(sh[aeiouáéíóúlnr]) $1 0 <[ANV].*> <&>
^for(s[^aeiouáéíóúlnrh]) $1 0 <[ANV].*> <&>
^fo-([aeiouáéíóú]) $1 0 <[ANV].*> <&>
# ^fo([lnr]) BELOW
^fo([bcdfgmpt]h) $1 0 <[ANV].*> <&>
^fo(sh[aeiouáéíóúlnr]) $1 0 <[ANV].*> <&>
^fo(s[^aeiouáéíóúlnrh]) $1 0 <[ANV].*> <&>
^fo-(..) $1 1 <[ANV].*> <&>
^frith([^bcdfgmpst-]) $1 0 <[ANV].*> <&>
^fri(t[^h]) $1 0 <[ANV].*> <&>
^frith([bcdfgmp]h) $1 0 <[ANV].*> <&>
^frith(sh[aeiouáéíóúlnr]) $1 0 <[ANV].*> <&>
^frith(s[^aeiouáéíóúlnrh]) $1 0 <[ANV].*> <&>
^h?iar([^bcdfgmprst-]) $1 0 <[ANV].*> <&>
^h?iar-(r) $1 0 <[ANV].*> <&>
^h?iar([bcdfgmpt]h) $1 0 <[ANV].*> <&>
^h?iar(sh[aeiouáéíóúlnr]) $1 0 <[ANV].*> <&>
^h?iar(s[^aeiouáéíóúlnrh]) $1 0 <[ANV].*> <&>
# ilsiamsa, ilsiollach, etc. no special case
^h?il([^bcfgmp-]..) $1 0 <[AN].*> <&>
^h?il([bcfgmp]h.) $1 0 <[AN].*> <&>
^h?im([^bcdfghmpst-]..) $1 0 <[ANV].*> <&>
^h?im([bcdfgmpt]h.) $1 0 <[ANV].*> <&>
^h?im(sh[aeiouáéíóúlnr]) $1 0 <[ANV].*> <&>
^h?im(s[^aeiouáéíóúlnrh].) $1 0 <[ANV].*> <&>
# ^in([^bcfgmp-]) BELOW
^h?in-(n) $1 0 <[AN].*> <&>
^h?in([bcfgmp]h..) $1 0 <[AN].*> <&>
^leas-(s) $1 0 <[AN].*> <&>
^leas([aeiouáéíóúd]..) $1 0 <[AN].*> <&>
^leas([bcfgm]h..) $1 0 <[AN].*> <&>
^mí-([aeiouáéíóú]) $1 0 <[ANV].*> <&>
^mí([aeiouáéíóú]..) $1 2 <[ANV].*> <&> # míúsáid
^mí([lnr]..) $1 0 <[ANV].*> <&>
^mí([bcdfgmpt]h.) $1 0 <[ANV].*> <&>
^mí(sh[aeiouáéíóúlnr]) $1 0 <[ANV].*> <&>
^mí(s[^aeiouáéíóúlnrh]) $1 0 <[ANV].*> <&>
^neamh-?([aeiouáéíóú]) $1 0 <[ANV].*> <&> # neamh-aistear but neamhaithne
^neamh([lnr]) $1 0 <[ANV].*> <&>
^neamh([bcdfgpt]h) $1 0 <[ANV].*> <&>
^neamh-(mh) $1 0 <[ANV].*> <&>
^neamh(sh[aeiouáéíóúlnr]) $1 0 <[ANV].*> <&>
^neamh(s[^aeiouáéíóúlnrh]) $1 0 <[ANV].*> <&>
^príomh([^bcdfgmpst-]) $1 0 <N.*> <&>
^príomh([bcdfgpt]h) $1 0 <N.*> <&>
^príomh-(mh) $1 0 <N.*> <&>
^príomh(sh[aeiouáéíóúlnr]) $1 0 <N.*> <&>
^príomh(s[^aeiouáéíóúlnrh]) $1 0 <N.*> <&>
^réamh([^bcdfgmpst-]) $1 0 <[ANV].*> <&>
^réamh([bcdfgpt]h) $1 0 <[ANV].*> <&>
^réamh-(mh) $1 0 <[ANV].*> <&>
^réamh(sh[aeiouáéíóúlnr]) $1 0 <[ANV].*> <&>
^réamh(s[^aeiouáéíóúlnrh]) $1 0 <[ANV].*> <&>
^rí([lnr]..) $1 0 <[ANV].*> <&>
^rí([bcdfgmpt]h.) $1 0 <[ANV].*> <&>
^rí(sh[aeiouáéíóúlnr]) $1 0 <[ANV].*> <&>
^rí(s[^aeiouáéíóúlnrh]) $1 0 <[ANV].*> <&>
^ró-([aeiouáéíóú]..) $1 0 <[AN].*> <&>
^ró([aeouáéíóú]..) $1 2 <[AN].*> <&> # not "i" since several actual words in Dinneen start with rói-
^ró([lnr]..) $1 0 <[AN].*> <&>
^ró([bcdfgmpt]h.) $1 0 <[AN].*> <&>
^ró(sh[aeiouáéíóúlnr]) $1 0 <[AN].*> <&>
^ró(s[^aeiouáéíóúlnrh]) $1 0 <[AN].*> <&>
^sain([^bcdfgmpst-]) $1 0 <[ANV].*> <&>
^sain([bcdfgmpt]h) $1 0 <[ANV].*> <&>
^sain(sh[aeiouáéíóúlnr]) $1 0 <[ANV].*> <&>
^sain(s[^aeiouáéíóúlnrh]) $1 0 <[ANV].*> <&>
^so-([aeiouáéíóú]..) $1 0 <[AN].*> <&>
^so([lnr]..) $1 0 <[AN].*> <&>
^so([bcdfgmpt]h.) $1 0 <[AN].*> <&>
^so(sh[aeiouáéíóúlnr].) $1 0 <[AN].*> <&>
^so(s[^aeiouáéíóúlnrh]..) $1 0 <[AN].*> <&>
^tras([^bcfghmnps-]) $1 0 <[ANV].*> <&> # leave "^trasn" alone
^tras([bcfgmp]h) $1 0 <[ANV].*> <&>
^tras-(s) $1 0 <[ANV].*> <&>
# next batch are non-words so really need to be here, but VERY infrequent
^(cil|gig|h?is|meig|micr|pic|teil)ea-?([^aeiouáéíóú-]+[aouáóú]) $2 0 <.+> <&>
^(cil|gig|h?is|meig|micr|pic|teil)i-?([^aeiouáéíóú-]+[eiéí]) $2 0 <.+> <&>
^(h?ant|fót|nan|par|pol|h?ultr)a-?([^aeiouáéíóú-]+[aouáóú]) $2 0 <.+> <&>
^(h?ant|fót|nan|par|pol|h?ultr)ai-?([^aeiouáéíóú-]+[eiéí]) $2 0 <.+> <&>
^(h?eachtar|freas|h?íos|neas|réalt|tob|h?uas|h?uath)-?([^-]) $2 0 <.+> <&>
#############################################################################
# Spelling standardizations, but ones where I'm less sure that they
# always apply, i.e. non-standard bits that are also commonly standard,
# e.g. -thaí$ which could be legit "torthaí" but also non-std "glórthaí"
# Because of this, safer to apply them after the prefix stripping, so
# e.g. "iarchoirpigh" comes out as MOIRF{choirpigh} instead of
# CAIGHDEAN{choirpeach}, and "fíorghanntanas" comes out as
# MOIRF{ghanntanas} instead of CAIGHDEAN{ghanntan}
chaí$ cha 1 <[ANF].*> <&>
thaí$ tha 1 <.+> <&> # glórthaí, spéarthaí
(.[aá])igh$ $1ch 1 <[NF].*> <&> # nb2 -aí genitives
(..)i[dg]h$ $1each 1 <[NF].*> <&> # same (e.g. stámhaillidh)
(..)aidh$ $1a 1 <.+> <&> # bhfágaidh
# past tenses
(...)eas$ $1 1 <V p="y" t="caite"> <&>
(..[^aeiouáéíóú])as$ $1 1 <V p="y" t="caite"> <&>
# rule for [^eéií] is above with other non-std prefixes, last resort here
# for words like cóimhéigean -> comhéigean
^cóimh([eéií]) comh$1 1 <.+> <&>
#############################################################################
# derivational morphology
([ao])chai?s$ $1ch 0 <A pl="n" gnt="n".*> <&>
acht$ ach 0 <A pl="n" gnt="n".*> <N pl="n" gnt="n" gnd="f">
([ao])chtaí$ $1cht 1 <N pl="n" gnt="n".*> <&>
íocht$ íoch 0 <A pl="n" gnt="n".*> <N pl="n" gnt="n" gnd="f">
úlachta?$ úil 0 <A pl="n" gnt="n".*> <&> # CO p.120
úchái?n$ ú 0 <N.*> <&>
# but actually non-standard if it's gs or pp; general uith -> aith below
uithe$ ú 0 <N.*> <&>
#############################################################################
# emphatic suffixes
(.[aouáóú])(san?|na)$ $1 0 <[NV].*> <&>
(.[eiéí])(sean|[sn]e)$ $1 0 <[NV].*> <&>
([aouáóú][^aeiouáéíóú-]*[^aeiouáéíóús-])san?$ $1 0 <[NV].*> <&>
([aouáóú][^aeiouáéíóú-]*[^aeiouáéíóún-])na$ $1 0 <[NV].*> <&>
([eiéí][^aeiouáéíóú-]*[^aeiouáéíóús-])se(an)?$ $1 0 <[NV].*> <&>
([eiéí][^aeiouáéíóú-]*[^aeiouáéíóún-])ne$ $1 0 <[NV].*> <&>
([aouáóú][^aeiouáéíóú-]*s)-san?$ $1 0 <[NV].*> <&>
([aouáóú][^aeiouáéíóú-]*n)-na$ $1 0 <[NV].*> <&>
([eiéí][^aeiouáéíóú-]*s)-se(an)?$ $1 0 <[NV].*> <&>
([eiéí][^aeiouáéíóú-]*n)-ne$ $1 0 <[NV].*> <&>
#############################################################################
# common spelling errors
# iri$ irí 2 <N.*> <&>
uil$ úil 2 <A.*> <&>
iúl$ iúil 2 <A.*> <&>
(.[^aeiouáéíóú])ala$ $1ála 2 <N.*> <&>
ula$ úla 2 <A.*> <&>
eail$ eáil 2 <.+> <&> # common missing accent
([^ae])oir$ $1óir 2 <.+> <&>
([^ae])ora$ $1óra 2 <.+> <&>
aio aío 2 <.+> <&> # ionadaiocht, srl
eor$ eoir 2 <N.*> <&>
ionn$ íonn 2 <V.*> <&>
itear$ ítear 2 <V.*> <&>
iteá$ íteá 2 <V.*> <&>
iodh$ íodh 2 <V.*> <&>
([^e])oidh$ $1óidh 2 <V.*> <&>
([^e])ofar$ $1ófar 2 <V.*> <&>
# next batch from ngramprobs.pl
u$ ú 2 <N.*> <&>
#chtai$ chtaí 2 <N.*> <&>
#ai$ aí 2 <.+> <&>
(...)i$ $1í 2 <.+> <&>
iu iú 2 <.+> <&>
no nó 2 <.+> <&>
# aio aío 2 <.+> <&>
#aioch(ta?)$ aíoch$1 2 <N.*> <&> # see prev.
io ío 2 <.+> <&>
leir léir 2 <.+> <&>
^(g?ch?)om([^haómnpr]..) $1omh$2 2 <.+> <&>
éá éa 2 <.+> <&> # held down alt too long; "éa" is only diphthong with "a" as second letter, and none with e/u
([áéóú])í $1i 2 <.+> <&>
íó ío 2 <.+> <&>
# places where the second vowel is the one with the fada are rarer - most of
# them are "aí", then "eá", "iú", "uí", then noise -- all but "iú" conflict
# with a resolution above; áí is, e.g., almost always a mistake for "ái" !
í([áú]) i$1 2 <.+> <&>
# i$ í 2 <.+> <&> # lotsa false compounds with this
#############################################################################
# rules with somewhat lower probability (e.g. fo-, must be after non-stnd
# verb endings to avoid "foluíonn" parsing as "fo+luíonn"
^do([lnr]..) $1 0 <[AN].*> <&>
^fo([lnr]..) $1 0 <[ANV].*> <&>
^h?in([^bcfgmp-]..) $1 0 <[AN].*> <&>
# these next ones work for proper names like "NGael"; otherwise
# they're caught by general "make everything lower" catch all that follows
^BP bP 1 <.+> <&>
^BHF bhF 1 <.+> <&>
^DT dT 1 <.+> <&>
^GC gC 1 <.+> <&>
^MB mB 1 <.+> <&>
^ND nD 1 <.+> <&>
^NG nG 1 <.+> <&>
^TS tS 1 <.+> <&>
# last resort for weird caps
([A-ZÁÉÍÓÚ]) \l$1 2 <.+> <&> # pRíOMhach
([eiéí][^aeiouáéíóú]+)íní?$ $1 0 <.+> <&> # diminutive
([aouáóú][^aeiouáéíóú]+)íní?$ $1 2 <.+> <&> # diminutive
([óo])igíní?$ $1g 0 <N.*> <&>
# the next rule handles a lot of stuff, including unwanted hyphens
# after prefixes (since the prefix rules will apply after recursing)
# and similarly unwanted hyphens before emphatic endings (CO p.126)
(.)- $1 1 <.+> <&> # ró-bheag / róbheag
# delenite, de-eclipse.
# Only helps when we recurse and find a standard unmutated prefix above!
# putting caps back Dec 08 for non-words like "tSíneach", "nIónach"
# which will only be found by stripping mutation and keeping cap
^b([Pp][^h]) $1 0 <[ANV].*> <&>
^bh([Ff][^h]) $1 0 <[ANV].*> <&>
^d([Tt][^h]) $1 0 <[ANV].*> <&>
^g([Cc][^h]) $1 0 <[ANV].*> <&>
^m([Bb][^h]) $1 0 <[ANV].*> <&>
^n([DdGg][^h]) $1 0 <[ANV].*> <&>
^t([Ss][^h]) $1 0 <[ANV].*> <&>
^([bcdfgmpBCDFGMP])h $1 0 <[ANVY].*> <&>
^([Tt])h([^s]) $1$2 0 <[ANVY].*> <&> # luathscéal danger
^([Ss])h([lnraeiouáéíóú]) $1$2 0 <[ANVY].*> <&>
^t'([AEIOUÁÉÍÓÚaeiouáéíóú]|[Ff]h) d'$1 1 <[ANVY].*> <&> # t'athair
^[bdm]'([AEIOUÁÉÍÓÚaeiouáéíóú]|[Ff]h) $1 0 <[ANVY].*> <&> # d'Éamonn too
^[nt]-([aeiouáéíóú]) $1 0 <[ANV].*> <&>
^[nt]([AEIOUÁÉÍÓÚ]) $1 0 <[ANV].*> <&>