Logo Lanfrica

selffins/somali

Domaine:

natural language processing

Type de record:

software
Créateur:
sel
Hôte:
Some very basic somali morphophonology on verbs/nouns using foma (homework) * Files ** `README.org` Files, usage, comments, list of underlying forms for roots, suffixes, and words; and conclusion. ** `somali.lexc` Generates the words with morpheme boundaries (before applying the FST.) ** `somali.xfst` The FST. Analyzes the surface forms, and generates the vocabulary. ** `surface.txt` The surface forms of the given data. ** `analyzed.txt` The output of the top-down application of the FST to `surface.txt`. * Usage To use, type ````foma -l somali.xfst``` in terminal. It applies the .xfst to `surface.txt` and generates a file `analyzed.txt`. * Comments ** General Replaced d͡ʒ with J and h̵ with H for convenience. ** `.lexc` and `.xfst` The roots for the nouns is largely the same as the +Sg surface forms. We hypothesize however, that some roots end in "m" but become "n" when syllable-final. The rules used, in prose and in order, are: *** Syllable-final m m becomes n when syllable final (precedes consonant or is word-final) **** Example: laam becomes laan. *** Vowel deletion for three syllables When we have the pattern VFVC^o (Vowel Fricative Vowel Non-Fricative o) or VrVP^aj (Vowel r Plosive o), we delete the second vowel. **** Example baɣal^o becomes baɣl^o. *** De-spirantization of syllable-final fricative When β, ð, ɣ or ʐ are syllable-final they get despirantized. Note how this rule is applied after Vowel deletion in VCVC. **** Example baɣl^o becomes baglo. *** Deletion of t t gets deleted when preceded by d, ɖ.x **** Example: kefed^ta becomes kefeda. *** ln to ll **** Example: qosol^naj becomes qosollaj. *** lt to sh **** Example: qosol^taj becomes qosoʃaj. * Underlying forms ** Suffixes *** Nouns |---------+-----| | +Sg | -∅ | | +Sg.Def | -ta | | +Pl | -o | |---------+-----| *** Verbs |-----------+------| | +3Sg.Masc | -aj | | +3Sg.Fem | -taj | | +1Pl.Past | -naj | |-----------+------| ** Roots *** Nouns Jið Jilin bað baɣal bil bohol daan daar dan feeʐ gaʕam gaβaɖ gees hoɣol irbað jir …