Chamorro/Transducer

Code: https://github.swarthmore.edu/ltarlin1/ling073-cha

Notes

• Currently the continuative tests are failing for verbs but succeeding for my one noun example. I believe that these were working before I added a few other tests in this section, so I may have accidentally influenced the previous tests. I still have to look into this further.
• I have not implemented the indefinite object marker prefix tests that require more than simply adding a prefix. For example, the verb 'tufok' becomes 'mamfok' with an indefinite object, so I need a rule to remove the first two letters of the stem when adding the prefix.
• Some of my plural subject marker tests are failing when there is a change in the location of the acute accent marking the stressed syllable. I am in the process of writing a rule that will hopefully fix this, but I havenʼt finished it yet.
• Some types of pluralization require reduplication or removing letters from the stem when adding the prefix--both of these I havenʼt accounted for yet.
• A few of my possession tests have a change in syllable stress and one has a vowel change (/i/ to /u/) that I havenʼt accounted for yet.

Analyser Evaluation

After first run, coverage is 3.90%.

By adding the following analyses, the coverage increased to 6.89%:

• siha<prn><px3pl> ↔ siha
• hit<prn><px1pl><inc> ↔ hit
• ma<prn><px3pl> ↔ ma

Missed words

Word	Occurences
i	87
na	74
ya	42
yan	30
y	27
gi	26
gui	24
a	24
ta	19
ni	19
Ya	19
si	19
ma	16
ha	15
g	14
u	12
o	12
Assan	10
h	9
ina	8

There are 55 stems in the transducer. Current coverage is 6.89% over the corpus.

Test results:

• cha.yaml = 30/51 passing
• commonwords.yaml = 2/20 passing

Generator Evaluation

Initial evaluation of morphological generation

• 30/51 analysis tests passing
• current coverage = 6.89%
• 30/62 generation tests passing