Mchampma1/final project
For my final project I expanded the adjectives in the western valley zapotec (zab) transducer.
Overview
Zapotec adjectives take two main forms, those which can be used as a predicate:
nahll cafee 'cold coffee'
and those that require a verb to have meaning
a bzhyag 'be drunk'
This required a new type of adjective sorting so that when analyzing and generating zab data the correct form of the adjective is generated. This reshaped the entries in the adjective category to be separated into predicative and modifying adjectives. The new form of the adjectives looks is now:
nga[pred] ngàaʼah[pred] # "green, raw" zagru[mod] zagrùu[mod] # "pretty"
Importantly, the double orthographies of the zab transducer are maintained. New adjectives, and the separation of the adjectives comes from Comparative constructions in Spanish and San Lucas Quiavini Zapotec. As well, all adjectives from "Cali Chiu" lesson 19 have been added to the transducer and some adjectives from the "talking Dictionary" have been added
Amount added
The current number of additional stems in the Lexd for zab is 31, with two orthographies for each adjective.
Evaluation
Below we will see the evaluation of the additions on a new corpus of adjective phrases from Comparative constructions in Spanish and San Lucas Quiavini Zapotec, and the talking dictionary, and a corpus of poems and sentences from the Ticha Textbook.
New Corpus
coverage: 48 / 63 (~0.76190476190476190476) remaining unknown forms: 15 ambiguity: 83 / 63 (~1.31746031746031746032)
All unknown forms in this small corpus are verbs and nouns which appear with adjectives in Comparative constructions in Spanish and San Lucas Quiavini Zapotec.
BxTP
coverage: 2411 / 3231 (~0.74620860414732281028) remaining unknown forms: 820 ambiguity: 3778 / 3231 (~1.16929743113587124729)
This corpus, like most of the available zapotec corpuses do not have extensive use of adjectives, which makes evaluation on existing corpora difficult.
