smn-phon-phonology

Alphabet

The Inari Saami letters

  • a á â ä b c d e f g h i j k l m n o õ p q r s t u v w x y z æ ä ø ö å č đ ŋ ŧ š ž
  • A Á Â Å B C D E F G H I J K L M N O Õ P Q R S T U V W X Y Z Æ Ä Ø Ö Å Č Đ Ŋ Ŧ Š Ž
  • %-
  • ' the hyphen as hyphen
  • ⎈ the ghost sign used in compounding, does not work in hfst

Accented letters, just in case

  • É Ó Ú Í À È Ò Ù Ì Ë Ü Ï Ê Ô Û Î Ã Ý þ Ð é ó ú í à è ò ù ì ë ü ï ê ô û î ã ý þ ð ß ª ß ç

Literal quotes and angles must be escaped (cf morpheme boundaries further down):

  • » « > <

Different archiphonemes

  • p4: p k4: k l4: l t4: t č4: č c4: c these are consonants that change in cg
  • i4: i this is the underlying i consonant
  • i2: i u2: u â2: â this is final i, u in contract verbs etc changing to e, o
  • i3: i this is invariant i
  • a5: a á5: á e5: e i5: i u5: u y5: y o5: o ä5: ä these do not change
  • c5: c č5: č d5: d h5: h j5: j l5: l m5: m ŋ5: ŋ r5: r t5: t z5: z these do not change
  • b6: b d6: d g6: g clitic consonants, jiemge, epke
  • j6: 0 a fake consonant, to get WG rules to function, e.g. puávui
  • i6: 0 a fake vowel, to get WG rules to function, e.g. kyehtlov (kyehti6lov)
  • '7: ' is the real apostroph

More abstract archiphonemes

  • %^RC: 0 = root cons., lengthened by ^CLEN - máán%RCu máánnun
  • %^RV: 0 = root vowel, lengthened by ^RLEN - ta%^RVvlu taavlu
  • %^SC: 0 = stem cons, %^SC: j, pluss lengthen after earlier shortening by %^CSH, see lyeme
  • %^SV: 0 = stem vowel, lengthened by ^SLEN u: uu - táálu%^SV tááloost
  • %^VO: 0 = vowel copy

Triggers

  • %^CLEN: 0 = Cons lengthening, with ^RC
  • %^CSH: 0 = Cons shortening, laaŋkku - laaŋkun
  • %^RVSH: 0 = Root vowel shortening
  • %^FCD: 0 = Final consonant deletion
  • %^FVD: 0 = Final vowel deletion
  • %^EA: 0 = Stem vowel: e and i to a/á in ill, and root vowel ä>á, e(e)>iä, ie>iä, ye>uá
  • %^EX: 0 = Stem vowel: i to â where it should have been á, this is Err/Orth only
  • %^RLEN: 0 = Root vowel lengthening
  • %^SLEN: 0 = Stem vowel lengthening
  • %^SVLOW: 0 = Stem vowel lowering â>á and u>o - árppu árppoin, Root vowel shortening when more than one cns in consonant centre and suffix starts with i
  • %^SVSH: 0 = Stem vowel shortening kappeer kaperân
  • %^WG: 0 = Weak grade trigger
  • %^ÁE: 0 = á to e in 3-syll, triggered by ^WG
  • %^ÁI: 0 = á to i in 3-syll, triggered by ^WG
  • %^VHIGH: 0 = High rootvow for verbs: e>iä, a>o, o>u, ää>áá, ee>iä, note also: ie>iä, ye>uá, uo>uá
  • %^VBACK: 0 = Back rootvow for verbs: ä(ä)>a(a), á>aa, uá>o, ye>uo
  • %^BLOCK: 0 = This symbol just to block otherwise triggering contexts
  • %^IUML: 0 = â to e in front of high suffixes

Trigger ordering in twolc and lexc, from left to right:

SV|ÁE|ÁI  EA|EX|SC  FCD|FVD  WG|CSH  CLEN  VHIGH|VBACK  RVSH|RLEN  SLEN|SVSH  SVLOW

Morpheme boundaries:

Here we define the suffix border > The other borders are not used yet, but still defined:

«
Derivational prefix
»
Derivational suffix
%<
Inflectional prefx
%>
Inflectional suffix
%^
(exceptional) soft hyphenation point and the symbol #, which is Word boundary for both lexicalised and dynamic compounds

End of alphabet definitions

Sets

  • Vow = a á â ä e i o u u2 y æ ø ö å i2 u2 â2 i3 a5 á5 i5 u5 y5 o5 ä5 i6 é ó ú í à è ò ù ì ë ü ï ê ô û î ã ý A Á Â Ä E I O U Y Æ Ø Ö Å É Ó Ú Í À È Ò Ú Ì Ë Ü Ï Ê Ô Û Î Ã Ý %^RV %^SV ;
  • RealVow = a á â ä e i o u y æ ø ö å y5 i2 i5 i3 i6 â2 é ó ú í à è ò ù ì ë ü ï ê ô û î ã ý A Á Â Ä E I O U Y Æ Ø Ö Å É Ó Ú Í À È Ò Ú Ì Ë Ü Ï Ê Ô Û Î Ã Ý ;
  • Cns = b c č d đ f g h j k l m n ŋ p q r s š t ŧ v w x z ž B C Č D Đ F G H J K L M N Ŋ P Q R S Š T Ŧ V W X Z Ž g2 k2 m2 n2 b3 d3 g3 m3 h3 p4 k4 r4 l4 i4 t4 c4 č4 h5 j5 ŋ5 t5 c5 l5 t5 d5 r5 z5 j6 %^RC %^SC ;
  • UnvoicedCns = c č f h k p s š t ;
  • Geminate = b d g p h t k c f z l m n ŋ r s š v ;
  • Sgm = Vow Cns ;
  • NotSuffixBorder = Vow %^FCD %FVD Cns %^CSH %^WG %^CLEN %^SLEN %^RLEN %^SVLOW %^RVSH %^ÁE %^ÁI %^EA %^EX %^SVSH ;
  • Triggers = %^FCD %^FVD %^WG %^CSH %^CLEN %^SLEN %^RLEN %^RVSH %^SVLOW %^ÁE %^EA %^EX %^ÁI %^SVSH %^VHIGH %^VBACK %^IUML ;
  • TriggersnotFCVD = %^WG %^CSH %^CLEN %^SLEN %^RLEN %^RVSH %^SVLOW %^ÁE %^EA %^EX %^ÁI %^SVSH %^VHIGH %^VBACK %^IUML ;
  • StemCns = b b3 d d3 g g2 g3 h h2 h3 j l m m2 m3 m8 n n8 n2 ŋ r s ;

Definition section

Introducing a WG right context

WeakGrade =  (RealVow:) [:RealVow |i6:]  (Cns:) ([%^SV:|%^ÁI:|%^ÁE:]) (%^FCD:) %^WG:0   ; 

As we look at more cases, the WeakGrade definition tends to be longer. We want to keep it short, but we want the code to work Thus, we document.

  • The second : RealVow is there for long consonant stems
  • The (Cns: ) is there for consonant stems

Rule section

Consonant rules

Consonant gradation

  • x:y consonant gradation
    WG gives x: x and x: y, when there is no consonant lengthening
  • va^RVh%^RCo^WG^CLEN>im
  • va0vvo00>im
  • Quality change of root consonant x:yy
    consonant lengthening after WG, changing h: v, p: v, c: s
  • t:đ, also tk4:đh gradation, part 1
  • jo^RVtk4â^WG^RLEN>
  • joođhâ00>
  • k4:h gradation for kk4, tk4, rkk4, lkk4
  • xx:yy gradation for xx4
    puáhhoin, sukk4á: suhháást
  • mä^RVtk4i^WG^RLEN>.#.
  • määđhi00>.#.
  • su^RVkk4á^SV^WG^CLEN^SLEN>st
  • su0hháá000>st
  • h:v gradation for hh
    já%^RVhhá%^SVl%^ÁI : jäävvil
  • kahhaan^CSH^SVSH>ân
  • ka0h0an00>ân
  • x4:z gradation for xx4
  • ru^RVtt4â^WG^RLEN>
  • ruu0đâ00>
  • haavvâr^CSH^SVSH>ist
  • ha00vâr00>ist
  • pottáák^SC^CSH^SVSH>ist
  • po0t0ákk00>ist
  • xy:zy gradation
    ht: vt, we use h
  • pihtâs^WG
  • pivtâs0
  • totk4os^WG
  • tođhos0
  • á^RVrppu^WG^RLEN>
  • áár0pu00>
  • á^RVrppu^WG^RVSH>ttáá
  • á0r0pu00>ttáá
  • riänttus^WG>
  • riän0tus0>
  • ku^RVrrâ^CSH^RLEN>n
  • kuu0râ00>n
  • so^RVllâ^CSH^RLEN>n
  • soo0lâ00>n

Consonant shortening

  • k:0 gradation for xkk4
    for lkk4, rkk4, vkk4
  • xyy:xy gradation A
    há%^RVi4ttás%^ÁI
  • č:0 in čč4 consonant gradation when CLEN
    when čč4
  • i^RVš^RCe^WG>
  • i0š0e0>
  • lu^RVčč4â^WG^RLEN^SVLOW>in
  • luu0já000>in
  • From i4j to j when WG and CSH
    e.g. sai4joos: sajoos
  • xx:x consonant gradation and shortening and x:x shortening
    weak grade or consonant shortening
  • vi^RVššâ^CSH^RVSH>vettee
  • vi00šâ00>vettee
  • la^RVs^RCe^SVtt^RLEN^SLEN>vettee
  • laas0ee0t00>vettee
  • cuozzâ^WG^RLEN>
  • cuo0zâ00>
  • vá^RVddu^CSH^RLEN>n
  • váá0du00>n
  • tt:t - final consonant and consonant gradation
    consonant gradation, and 3syll verbs in -tt lose t in front of cns-initial suff
  • kulluutt^SVSH>
  • kullu00t0>
  • las^RCe^SVtt^CLEN^SLEN>
  • lassee0t00>
  • nnj to nj shortening weak grade and consonant shortening
    special rule, e.g. konnjâl -> konjâlân puánnjá
  • Word final consonant deletion
    for vaskâm: vaaskâ
  • va^RVskâm^FCD^RLEN
  • vaaskâ000
  • njuolgâd^FCD^RVSH
  • njuolgâ000
  • Final m in stems, smm to sm
  • Wordfinal consonants part 1 e%^RVlâčč 1) 2)jurdâččmân: jurdâšmân ju%^RVrdâčč%^RLEN%>mân 3) šš>š in hárávušš>vetteđ
  • hárávušš>vetteđ
  • hárávu0š>vetteđ
  • konnjâl^CSH>ân
  • kon0jâl0>ân
  • suán^RCju^CLEN>n
  • suánnju0>n
  • haamit^SC^CSH^SVSH>ist
  • ha0mitt00>ist
  • su^RVkk4á^SV^WG^CLEN^SLEN>st
  • su0hháá000>st

Consonant lengthening

  • Lengthening of root consonant x:yy
    consonant lengthening after WG - h: vv, p: vv, c: ss
  • ka^RVss^RCa^SV^WG^CLEN^SLEN>st
  • ka0ss0aa000>st
  • va^RVh^RCe^SV^WG^CLEN^SLEN>st
  • va0vvee000>st
  • va^RVh^RCe^SV^CLEN^SLEN>n
  • va0hhee00>n
  • Lengthening of half-long root consonant x:xx
    riččá piiru piirrun niisu niissun
  • Lengthening of consonant h5
    this one does not fit into the general rule because of h5: h
  • put^RCe^SVst^CLEN^SLEN>
  • putteest00>
  • ke^RVp^RCidm^CLEN^RLEN^SLEN>ân
  • keeppidm000>ân
  • kieh^RCâ^WG^CLEN^SVLOW>in
  • kievvá000>in
  • i^RVšš^RCe^SV^WG^CLEN^SLEN>st
  • i0šš0ee000>st
  • kol^RCe^CLEN>
  • kolle0>
  • kiel^RCâ^CLEN>n
  • kiellâ0>n
  • č:ij part 1
  • Lengthening root consonant j:ij 1
    for saje: saijeen
  • č:ij and j:ij part 2
  • sa^RVj^RCe^SV^CLEN^SLEN>n
  • sa0ijee00>n
  • pááj^RCá^CLEN>n
  • pááijá0>n
  • vaj^RCe^SVd^CLEN^SLEN>
  • vaijeed00>
  • nj to nnj consonant lengthening
    special rule, suánju : suánnjun for Par and Ess
  • Consonant lengthening after xx:x and xy:zy weak grade with x4 on upper side
    k4, č, p4 kukse: kuvssijn
  • Consonant lengthening xy:xyy

Stem consonant and suffix consonant

  • Assimilation of consonants t d š delet t in front of -ttáá, avoid -tttáá
  • haamit>ttáá
  • haami0>ttáá
  • Wordfinal consonants part 2 e%^RVlâčč cummâlâhč: cummâlâhš
  • Final consonant x:xx
    pottáák potákkân
  • kahhaan^SC^CSH^RVSH>ist
  • ka0ha0nj00>ist

Vowel rules

First syllable vowel shortening

  • koonnjâl^CSH>ân
  • ko0n0jâl0>ân
  • kappeer^CSH^SVSH>ist
  • ka0p0er00>ist
  • čäällim^CSH>ist
  • čä00lim0>ist
  • muusik^SC^CSH^RVSH>ist
  • mu0sikk00>ist
  • aarrân^CSH>ân
  • a00rân0>ân
  • kiirrâs^RVSH>ân
  • ki0rrâs0>ân
  • Vowel shortening VV:V
    ráávhu- : rávhoid, käähir%^CSH%>id kuuđiâ%^RVSH%>žân
  • Long vowel shortening
    (example?), no ^CSH, since that gives Vy: 0
  • e:0 in passive
  • mááh^RCu^WG^CLEN^SVLOW>in
  • má0vvo000>in
  • mool^RCâ^WG^RVSH>ttáá
  • mo0l0â00>ttáá
  • e:0 in passive
    special rule for getting ej̣ittáá instead of ej̣ittáá.
  • Underlying long vowel shortening iä:e and ää:ä or a
    čääh%^RCu2%^CLEN%^VBACK%>um

3st stems viäskár veskirin, časkes časkes ča%^RVskas%^ÁE viäskár%^ÁI%^WG%^RVSH

  • kaavpug^CSH^RVSH^SVSH>ân
  • ka0vpug000>ân
  • mool^RCâ^WG>
  • mool0â0>
  • käähir^WG>
  • käävir0>
  • juáh^RCu^WG^RVSH>ttáá
  • juáv0u00>ttáá

First syllable vowel lengthening

  • Root vowel lenghtening in many stems
    no rootvowel length if stem vowel lengthens and the consonantcentre consists of two or more consonants
  • kä^RVvnâst^WG^VBACK^RLEN>
  • kaavnâst000>
  • le^RVkk4â^WG^RLEN>
  • lee0hâ00>
  • nju^RVne^SV^WG^RLEN^SLEN>st
  • njuunee000>st
  • Root vowel o5:u
  • Root vowel ä:ää lenghtening and diphthongisation e:iä
    special rule, for avoiding conflict spä%^RVi4lih%^WG%^RLEN,
  • Root vowel o lengthening
    alge : oolgijn
  • a^RVlge^WG^RLEN>id
  • oolgi00>jd
  • va^RVh^RCu^CSH^RLEN>
  • vooh0u00>
  • Root vowel u lengthening with vowel change before suffix i
    alge : oolgijn
  • lo^RVh^RCâ^RLEN>đ
  • luuh0â0>đ
  • mu^RVšti^WG^VHIGH^RLEN>j
  • muušti000>j
  • lo^RVh^RCá^WG^VHIGH^RLEN>im
  • luuv0á000>im
  • ku^RVlgâ^WG^VHIGH^RLEN>i3
  • kuulgâ000>i

First syllable vowel change

a rules

  • Root vowel change a:o before suffix i and for contract verbs
    alge : olgijn
  • Root vowel o:u for contract verbs and Root vowel change o:u before suffix i
  • tohhi2>đ
  • tuhhi>đ
  • lo^RVh^RCá^WG^VHIGH^RLEN>im
  • luuv0á000>im

ä rules

  • Monophthongisation rule iä:ee 1
    1)
  • iä:e rule 1 and second syllable Fleeting vowel before suffix
    Note: many rules in one
  • iä:e and iä:ee rule 2 and Diphthongisation i5ä to ie
  • siämmu2>đ>
  • s0emmu>đ>
  • Different rules for ä:a, first vowel (or only vowel)
    čääci: čassijn, läbži: labžijn lään^RCi%>id
  • njääh^RCi^WG^VBACK>i3
  • njaav0i00>i
  • ää to aa, second vowel
    á: o in pairs like kuátij -> koođij, säämi -> saamij čäällu%^VBACK%>m lään^RCi%>id
  • ää to áá and á0 in Illative and Sg3
    ... for säämi > sáámán
  • sääm^RCi^EA>n
  • sáám0á0>n
  • njääh^RCo^CLEN^VBACK^RVSH>on
  • njá0hho000>on
  • äRV to áá in Illative and Sg3
    ... for täsni -> táásnán
  • tä^RVsni^EA^CSH^RLEN>n
  • táásná000>n
  • Vowel lengthening äRV to aa
    äigi aaigijd
  • sälkki2>đ>
  • salkki>đ>
  • sääm^RCi^WG>in
  • saam0i0>jn
  • sääm^RCi^WG>id
  • saam0i0>jd
  • stäävi2>đ>
  • staavi>đ>
  • stäävi2>j>
  • staavi>j>
  • stäävi2>i2>
  • staavi>i>
  • stäävi2>jeh>
  • stääve>jeh>
  • kä%^RVvnâ^WG>st^RLEN>
  • kaavnâ0>st0>

á

  • Root vowel change for á to ä with i in second syllable
    päävir , já%^RVhhá%^SVl%^ÁI%^WG%^RLEN>
  • páápp4ár^ÁI^WG>
  • pää0vir00>
  • Root and stem vowel á:a for verbs
  • pá^RVkk4u2>m
  • pa0kku>m

á and uá rules

  • Monophthongisation rule uá:o0 part 2:Vowel shortening for uá
    for uábbi -> obbijn čuál%^RCi%^WG%>in tuái4jâl%^CSH%^VBACK%^RVSH%>âm
  • Monophthongisation rule u:o and root vowel change u:o, special rule for olmooš and contract verbs
  • uáinu2>đ
  • o0inu>đ
  • Diphthong rule uá:uo and uá:oo, part 2
  • Diphthongisation e to iä, part 1
  • Root vowel aa:áá in C-stems with stemvowel u, part 1
    ahhu: áhu
  • Root vowel aa:áá in C-stems with stemvowel u, part 2
    ahhu: ááhu
  • ie to iä in Illative and Sg3
    kiem´ni -> kiämnán, veerdi : viärdán peerru2%^CLEN%^VHIGH%^RVSH%>o
  • Diphthongisation uo:uá
  • ye to uá and uo, part 1
    Part 1 of ye to uá for kyeli : kuálán
  • ye to uá, part 2
    Part 2 of ye to uá for kyeli : kuálán
  • kyel^RCi3^EA^CLEN>m
  • kuállá00>m
  • ye to uo, part 2
    Part 2 of ye to uá for kyeli : kuolijn
  • uá to ye and uo to ye, part 1
    Part 1 luámmán : lyeme čuággás%^ÁI%^WG%^RLEN%>
  • vuolle^CLEN>en
  • vyelle0>en
  • uá to ye, part 2
    Part 2 luámmán : lyeme

Second syllable rules

  • Second syllable vowel change á to e or i in 3syll stems
    puttás- > putes, triggers are ^ÁE and ^ÁI já%^RVhhá%^SVl%^ÁI%^WG%^RLEN%> pu5árrás%^ÁI%^WG%>umos
  • puttás^ÁE^WG>
  • pu0tes00>
  • Stemvowel shortening áá and aa to e or i in 3syll
    puttás- > putes, triggers are ^ÁE and ^ÁI
  • nahhaar^ÁE^WG>
  • na0v0er00>
  • Stem vowel e to a for a root vowel.
    alge to aalgan
  • a^RVlge^EA^CSH^RLEN>n
  • aalga000>n
  • va^RVh^RCe^SV^EA^CLEN^SLEN>m
  • va0hhaa000>m
  • Stem vowel e to á for other root vowel
    if e stem
  • Stem vowel i to á for other root vowel.
    e.g. säämi > Ill sáámán
  • Stem vowel i to â for other root vowel, as a common error.
    e.g. säämi > Ill Err/Orth sáámân
  • Stem vowel i to e in front of j
  • Stem vowel â to a in Pl Nom -h
    silbâ to siilbah %^WG%^RLEN%>hlove%>h
  • so^RVllâ^WG^RLEN>h
  • soo0la00>h
  • Stem vowel â > á change before i
    sollâ to sooláid
  • so^RVllâ^WG^RLEN^SVLOW>in
  • soo0lá000>in
  • Stem vowel i3:0
  • Stem vowel u:o change before suffix
    2)
  • Stem vowel u2:o change before suffix
  • uáinu>o
  • uáino>o
  • Second vowel shortening
    eemeed : emedân, savvoon : savo
  • eemeed^CSH^SVSH>ân
  • 0em0ed00>ân
  • kappeer^CSH^SVSH>ân
  • ka0p0er00>ân
  • Stem vowel:0 in front of â
    aiguu: aaigâ
  • Stem â deleting
    kandâ: kandii

Suffix vowel rules

  • Suffix vowel lengthening
    long á in kissáást, nuorâ : nuorááin, hyeni: hyeneest
  • ki^RVssá^SV^WG^CLEN^SLEN>st
  • ki0ssáá000>st
  • mi^RVllo^SV^WG^CLEN^SLEN>st
  • mi0lloo000>st
  • huksi^SVm^SLEN>ist
  • huksiim0>ist
  • táál^RCu^SV^WG^RVSH^SLEN^SVLOW>st
  • táál0oo0000>st
  • Suffix i5:j for lexicon APINA
    salijn
  • Suffix i5:0 and i5ä:e
    timotein oboen, not Chariliein
  • Suffix i:j for i and e-stems
  • Vowel copy in suffix
    for imperative Sg3, at least

Suffix consonant rules

Other rules

  • Word final vowel deletion
    for vaskâm: vaaskâ
  • sollâ^FVD>ii
  • soll00>ii
  • Potential vowel shortening for contract verbs
    -áážep becomes -áš etc.
  • skäi4nâ^VBACK>đ
  • skainâ0>đ
  • skä^RVi4nâ^VBACK>đ
  • ska0inâ0>đ
  • hu^RVm^RCo^SVb^WG^RLEN^SLEN^RLEN^SLEN
  • huum0oob00000
  • hu^RVm^RCo^SVb^RLEN^SLEN^RVSH>ân
  • hu0m0o0b000>ân
  • hu^RVm^RCo^SVb^RVSH>ist
  • hu0m0o0b0>ist