North Sami NLP Grammar

Finite state and Constraint Grammar based analysers, proofing tools and other resources

View the project on GitHub giellalt/lang-sme

Page Content

sme fst guide

How to add new words to lexc files:

Adding gieđahit verb å hantere

Search for -hit verbs

gieđahit:gieđah MUITAL_TV ;

Adding

diskehallat - diskvalifisere (passive meaning)

diskehallat+OLang/NOB:diskehalla RAIMMAHALLA_IV ; ! ^LOAN

Adding Nay Pyi Taw

Search google Hovedstaden i Myanmar (Burma) search for -aw in smi-propernouns.lexc

Saginaw+OLang/UND:Saginaw BERN-plc ;
Nay% Pyi% Taw+MWE+OLang/UND:Nay% Pyi% Taw BERN-plc ;

Adding Frivilighet Norge

propernoun in smi-propernouns.lexc

Frivillighet% Norge+MWE+CmpNP/First+OLang/NOB:Frivillighet% Norge ACCRA-org ;

Adding Play-boy sub to Playboy

is a propernoun in smi (smi-propernouns.lexc) since there is no -boy I search for -y

Roy+CmpNP/None+OLang/UND:Roy BERN-mal ; but Playboy should be -org or -obj so better:

FairPlay+CmpNP/First+OLang/ENG:FairPlay BERN-obj ;
Playboy+Err/Orth+CmpNP/First+OLang/ENG:Play-boy BERN-obj ; !SUB

Adding propernoun Slagen

Google first

like Slaatten+OLang/UND:Slaatten9 LONDON-sur ; but since it is a place it needs to be LONDON-plc

Adding propernoun Tønsberg Blad

to shared-smi Sem/Org

Tønsberg% Blad+MWE+CmpNP/First+OLang/NOB:Tønsberg% Blad9 BERN-org ;

Adding last and first names

Mollekleiv - last name same ending as Hynnekleiv+OLang/NOB:Hynne^kleiv BERN-sur ;

Mollekleiv+OLang/NOB:Molle^kleiv BERN-sur ;
Butler+OLang/ENG:Butler LONDON-sur ;
Hoge+OLang/NOB:Hoge ACCRA-sur ;

Adding Nouns - perfomativitehtateoriija

ráŋggáštanriekteteoriija+N+Sem/Dummytag:ráŋggáštan#riekte#te^ori IIJA ;

perfomativitehtateoriija+N+Sem/Prod-cogn:perfomativitehta#te^ori IIJA ;

Adding Iežasnájadeapmi

is a noun: try to find another one with the same ending

AWG-lágideapmi+N+CmpN/SgN+CmpNP/First:AWG-lágid EAPMI_default_sem ;

but this one has a hyphen, those are special, so try to find one without

beassášávvudeapmi+N+CmpN/SgN+Sem/Event:beassáš#ávvud EAPMI_lex_sem ;
biebmoguollešaddadeapmi+N+CmpN/SgN:biebmo#guolle#šaddad EAPMI_default_sem ;
vuorbádeapmi+N+CmpN/SgN+CmpNP/First:vuorbád EAPMI_default_sem ;

how to know which one: =EAPMI_default_sem= gets a default semantic tag EAPMI_lex_sem gets a manual semantic tag for example Sem/Event in this case

means self-harm so it should be default_sem

  1. add at some place in nouns.lexc, if possible alphabetically, but not necessary, and put first letter in lower case
  2. copy ending from a similar entry: +N+CmpN/SgN+CmpNP/First:vuorbád EAPMI_default_sem ;
  3. switch the part after the colon and look at which letter it stops, so “iežasnájad”
  4. remove +CmpNP/First - iežasnájadeapmi+N+CmpN/SgN:iežasnájad EAPMI_default_sem ;
  5. if it is a compound then put a hashtag inbetween on the part after the comma: iežasnájadeapmi+N+CmpN/SgN:iežas#nájad EAPMI_default_sem ;
  6. run make -j in lang-sme
  7. test word - echo iežasnájadeapmi hfst-lookup -q src/analyser-gt-desc.hfstol

+CmpNP/First means: it can only be the first part of a compound, if not we put a spelling error - this is put if it is easily confused with another compound that is more common We also use it if the word is an MWE (Sámi% Dáiddaguovddáš+MWE+CmpNP/First:Sámi% Dáidda#guovddáž LONDON-org ;)

next word:

Adding noun siskkáldasdálkkodeapmi

  1. find the longest possible match (that is also a compound), so not only deapmi, but dálkkodeapmi –+N+CmpN/SgN:ele#rávdnje#dálkkod EAPMI_default_sem ;
  2. replace ele#rávdnje by siskkáldas: siskkáldasdálkkodeapmi+N+CmpN/SgN:siskkáldas#dálkkod EAPMI_default_sem ;
  3. DONE

Adding propernoun Árbediehtoguovddáš

it’s a propernoun since it is a sme-specific propernoun which would get translated in the other Sámi languages it goes to lang-sme/src/fst/stems/sme-propernouns.lexc (instead of shared-smi)

  1. look for the longest possible match meaning if we have a compound, let’s try to find a compound - diehtoguovddáš NO guovddáš: 
    Davvi% álbmogiid% guovddáš+MWE+CmpNP/First:Davvi% álbmogiid% guovddáž LONDON-org ;
Sámi% Dáiddaguovddáš+MWE+CmpNP/First:Sámi% Dáidda#guovddáž LONDON-org ;
  2. Add entry + tags from the match - Árbediehtoguovddáš+MWE+CmpNP/First:Sámi% Dáidda#guovddáž LONDON-org ;
  3. Change part after the colon and add hashtags if there are more parts of a compound - Árbediehtoguovddáš+MWE+CmpNP/First:Árbe#diehto#guovddáž LONDON-org ;
  4. and then remove +MWE since it’s only one word
  5. remove +CmpNP/First

Adding buotagat

It’s an adjective:

  1. search for a compound with -agat guhkesagat+A+Sem/Dummytag:guhkes#ag AGAdj ; CAREFUL (don’t confuse) with similar -agat words where the consonant before is part of the word, like - lagat+A+Sem/Dummytag+Gram/Comp:laga OVDDIT ;
  2. add entry with tags of guhkesagat
  3. replace parts after the colon:

These are tags that say that the entry can be the first part of the compound, it can either be in nominative singular and genitive plural.

+CmpN/SgN — could be buotagatsearvi +CmpN/PlG — could be buotagagiidsearvi (this is used in combination with Sem/Hum tag)

be aware of morphophonological processes: before the hashtag:

Adding unnibuš

It’s an adjective, but not yet in the normative lexicon

unnibuš unni+A+Der/Comp+A+Der/Dimin+A+Attr 0,000000
unnibuš unni+A+Der/Comp+A+Der/Dimin+A+Sg+Nom 0,000000
unnibuš unni+A+Err/Orth+Der/Comp+A+Der/Dimin+A+Attr 0,000000
unnibuš unni+A+Err/Orth+Der/Comp+A+Der/Dimin+A+Sg+Nom 0,000000

search for the ending -buš: stuorebuš+A+Sem/Hum:stuorebužž STUORIBUS ; exchange parts of it unnibuš+A+Sem/Hum:unnibužž STUORIBUS ;

Adding “Middagsfjället”

it is a propernoun for all Sámi languages open shared-smi/src/fst/stems/smi-propernouns.lexc search for fjället find: Borkafjället+OLang/UND:Borka^fjället LONDON-LOAN-plc ; adapt: Middagsfjället+OLang/UND:Middags^fjället LONDON-plc ;

Adding buotagat “Jensenius”

is a surname open shared-smi/src/fst/stems/smi-propernouns.lexc search for the ending -ius Iskanius+OLang/UND:Iskani^us BERN-sur ;

Adding buotagat “Rama Cay”

what on earth is that? - google knows an island in Nicaragua There is nothing similar in the lexicon The following continuation lexica mean the following: BERN-plc — Bernas, Bernii LONDON-plc — Londonis, Londonii ACCRA-plc — Accras, Accrai

search for “ay” – Bay+CmpNP/None+OLang/UND:Bay BERN-plc ; add +MWE change +CmpNP/None to +CmpNP/First

Adding propernoun Sziget

google - Hungarian festival

Sziget+OLang/UND:Sziget9 LONDON-org ;
``



## what does the number 9 mean?

bargiidbellodatpolitihkar+v1+N+CmpN/SgN+CmpN/SgG+CmpN/PlG+OLang/NOB+Sem/Hum:bargiid9#bellodat#politihkkar MATTAR ;

bargiid#bellodat#politihkkar MATTAR ; slamlaguna+N+CmpN/SgN+CmpN/SgG+CmpN/PlG+Sem/Plc:slam9#laguna GOAHTI-A ;

eaŋkilváikkuhangaskaoapmi+v5+N+OLang/NOB+Sem/Dummytag:eaŋkal#váikkuhan#gask9#oapmi GOAHTI-I ;


9 prevents that 
- bargiid- changes to bargiit-
- slam- changes to slan- (slanlaguna)
- en- changes to something else
- gask- changes to gas-

- vowel + d/m/k/s/h
- some continuation lexicons are an exception - Sotaniemi+OLang/FIN:Sota^niem NIEMI ;
  

## What does ^ mean?
`Playboy+CmpNP/First+OLang/ENG:Play^boy BERN-obj ;'

It's a soft hyphen for names as opposed to hard hyphens # in anything else

## What does º mean?

billu+N+Sem/Dummytag:bilºlu GOAHTI-U ; `` it means that genitive cannot be “bilu” - so it’s third grade/second grade

What kind of languages are there in propernouns?

What does +CmpNP/First mean?

We add it always after +MWE it means that it can only be the first part of a compound, e.g. “Frivillighet Norge-organisašuvdna”

Meaning of different continuation lexica

hit-verbs