Divvun & Giellatekno - open source grammars for Sámi and other languages

North Sámi compounding

This file governs prefixing and compounding, with the following lexica and pointers. All lexica and lexicon entries are documented.

LEXICON Prefixes = lexicon for adding *eahpe and pointing to N, A, V

LEXICON R = lexicon which is pointed to from affixes files. Here the strings get flags to control compounding (@P.CmpFrst.FALSE@ etc.) and are redirected to RAlmostReal.

LEXICON RAlmostReal = lexicon pointed to from R (where flags are added) and pointing to RrealAfterCmpNFlags and (with +Cmp tag) to MiddleNouns. lexicalising the 3-part compounds, with the tag ShCmp. It has two entries:

• Just pointing directly to RrealAfterCmpNFlags
• Adding +Cmp#:∑# and pointing to **MiddleNouns. These nouns should not return to themselves, to avoid -jotjotjot- They thus point directly to Rreal.

LEXICON Rreal = This is the former R lexicon, renamed to avoid the MiddleNouns loop. The string gets flags like for R, and directed to RrealAfterCmpNFlags.

• @P.CmpFrst.FALSE@ and other flags to control compounding

LEXICON RrealAfterCmpNFlags = This was also part of the former R lexicon, here renamed to avoid the MiddleNouns loop. Here it gets flags ensuring the result is N+N.

• N+N is the normal case.
• N+(V to N) ensured by Flag diacr restricting to V>N.
• N+(A to N) A needs a N tag later in the derivation
• Then 3 cases (points to N, V, A) add a hyphen, so Sem-julggaštus and maana-gåetie are allowed.
• Then 3 cases (to N, V, A) add a SOFT hyphen, to make it possible to analyse certain texts from printing houses and newspapers.
• to Acronym, maana-tv, “lomme-cd-spelar”
• to Lahka,
• to CmpNumeral, maana-123
• to ProperNoun, as the 2nd part of compounds for non-hyph. words. viessu-London goes through here.
• To words requiring hyphens, like -tv- and -cd-
• To ENDLEX, to take care of Oahppo- ja dutkandept

LEXICON RHyph = Recursive lexicon from all classes REQUIRING a hyphen to follow.

• Add Flags to control compounding, go to RHyphTags

LEXICON RHyphTags = adds +Cmp/Hyph and +Cmp, and then - on lower side.

• To Noun, the normal case.
• To HyphNouns, for nouns requiring hyphens, like -tv- and -cd-
• To Verb via flag diacr declares that the compound
• To A, needs a N tag later in the derivation
• To Acronym, like maana-tv, “lomme-cd-spelar”
• to Lahka,
• to CmpNumeral, NRK-2 etc.
• Proper nouns as the 2nd part of compounds for hyph-words. London-Hull is covered here, whereas Hull-viessu is covered by RHyph + Noun.
• To ENDLEX to take care of Oahpo- ja dutkandept - want this in speller

LEXICON RNum = For Num Cmp Noun, vi vil ikke ha Num Cmp Num

• Gives +Cmp/Hyph+Cmp and points to Noun

LEXICON Rnoun = the lexicon has two entries:

• either adds > and goes to the compound lexicon Rreal
• or goes to ENDLEX as Kárášjot, independent (sub) word, with +Err/Orth

LEXICON RProp = lexicon pointed to from propernouns, and containing 3 entries

• Flags to control compounding and to RPropTags
• nammasaš, points to DER-SAS
• nammasaš, points to AHKASAS, for MT

LEXICON RPropTags = A special lexicon for handling proper noun compounding without hyphens. Two entries:

• @C.CmpHyph@ RHyphTags ;: This is the regular case, giving hyphens to compounds
• @D.CmpHyph.TRUE@@U.CmpHyph.FALSE@+Err/Orth+Cmp/NoHyph+Cmp#:@D.CmpHyph.TRUE@@U.CmpHyph.FALSE@∑# Noun ;: This is the special case, going directly to nouns (not to NounRoot, as that would allow compounding with words explicitly coded to disallow such compounds)

LEXICON flagON-R = turns NeedsVowRed on:

• adds @U.NeedsVowRed.ON@ and directs to R

LEXICON flagOFF-R = turns NeedsVowRed off:

• adds @U.NeedsVowRed.OFF@ and directs to R

---

This (part of) documentation was generated from src/fst/morphology/compounding.lexc