Gulf Arabic (afb)

This repo contains the inflection tables for Gulf Arabic (ISO 639-3 afb)

Contents

• afb: entries based on lemmas that appear in the Annotated Gumar Corpus.
• afb.args: a UniMorph 4.0 compatible verion of afb
• afb.gloss: English glosses for the lemmas in afb.
• README.md: this file.

afb

Generation of the lemma inflections

• The inflections of most of the verb lemmas were generated through the CamelTools morphological generator component (demo, API) (Obeid et al., 2020). The morphological database used is CALIMA-GLF (Khalifa et al., 2017).
• The forms for the all nominal lemmas and some verbs are what appear in the Annotated Gumar Corpus (Khalifa et al., 2018). Therefore, the paradigms for them might not be complete. Additionally, to eliminate noisy entries arising from possible gold errors we used Morph/POS statistics from the same corpus to eliminate incorrect forms as much as possible.
• The POS and morphological features are then mapped to UniMorph according to the current schema (Sylak-Glassman 2016).
• The core POS of the lemmas are Verbs, Nouns, and Adjectives.
• The total number of lemmas is 6,707, with the following POS distribution:
  • V: 2,183 (32.6%) lemmas
  • N: 3,003 (44.8%) lemmas
  • ADJ: 1,520 (22.7%) lemmas

Source

• The Annotated Gumar Corpus (Khalifa et al., 2018). The corpus can be found here.

Notes on Tokenization and Diacritization

• Clitics were not included or marked in the inflection tables. The only clitic included is the determiner Al+ in order to be consistent with the other Arabic varieties in UniMorph.
• All the lemmas are diacritized. However, only the verb forms coming from CALIMA-GLF are diacritized. Removing all the diacritics is straightforward and can be done through a simple regex. Alternatively, CamelTools provides a dediacritization utility: an API and a CLI.

Notes on POS decisions

• All nominals with Al+ will be tagged with DEF for definiteness. All nominals without Al+ will be repeated twice: once as INDF and once as PSSD. That is because in most cases possession marking is not overt due to the orthography.
• All verbs are by default in the active voice.

Annotators

Salam Khalifa and Nizar Habash (CAMeL Lab @ NYU Abu Dhabi)

Paradigm Samples

The complete inflection table for the noun lemma بَركَن 'park (a vehicle)'

بَركَن	بَركَنتَوا	V;PFV;PL;2
بَركَن	تبَركِن	V;IPFV;FEM;SG;3
بَركَن	بَركَنَوا	V;PFV;PL;3
بَركَن	بَركِنَوا	V;IMP;PL;2
بَركَن	بَركَن	V;PFV;MASC;SG;3
بَركَن	بَركَنت	V;PFV;MASC;SG;2
بَركَن	بَركَنت	V;PFV;SG;1
بَركَن	بَركَنَّا	V;PFV;PL;1
بَركَن	بَركَنَت	V;PFV;FEM;SG;3
بَركَن	تبَركِنُون	V;IPFV;PL;2
بَركَن	اَبَركِن	V;IPFV;SG;1
بَركَن	تبَركِنِين	V;IPFV;FEM;SG;2
بَركَن	يبَركِنُون	V;IPFV;PL;3
بَركَن	بَركَنتِي	V;PFV;FEM;SG;2
بَركَن	بَركِن	V;MASC;IMP;SG;2
بَركَن	يبَركِن	V;IPFV;MASC;SG;3
بَركَن	تبَركِن	V;IPFV;MASC;SG;2
بَركَن	نبَركِن	V;IPFV;PL;1
بَركَن	بَركِنِي	V;FEM;IMP;SG;2

The complete inflection table for the noun lemma سِيّارَة 'car'

سِيّارَة	السيارة	N;DEF;FEM;SG
سِيّارَة	سياير	N;INDF;FEM;PL
سِيّارَة	سيار	N;INDF;FEM;SG
سِيّارَة	سيار	N;FEM;SG;PSSD
سِيّارَة	سيارتين	N;INDF;FEM;DU
سِيّارَة	السياير	N;DEF;FEM;PL
سِيّارَة	سياير	N;FEM;PL;PSSD
سِيّارَة	سيارتين	N;FEM;DU;PSSD
سِيّارَة	السيارتين	N;DEF;FEM;DU

License

• Creative Commons Attribution-ShareAlike 3.0 Unported (CC BY-SA 3.0)