Difference between revisions of "Warlpiri/Midterm"
From LING073
(→Challenges) |
|||
| (3 intermediate revisions by the same user not shown) | |||
| Line 15: | Line 15: | ||
==Evaluation== | ==Evaluation== | ||
*Currently getting about 40% coverage of my corpus. I increased this number dramatically this week by adding punctuation, because in Warlpiri, the suffixes are often added with hyphens, so most words contain hyphens. | *Currently getting about 40% coverage of my corpus. I increased this number dramatically this week by adding punctuation, because in Warlpiri, the suffixes are often added with hyphens, so most words contain hyphens. | ||
| − | *Total number of tokens: 1416 and total number of analyses: 1574 | + | *Total number of tokens: 1416 and total number of analyses: 1574 (which is puts my measure of ambiguity at 1.1) which means that I have some ambiguity to sort through. |
==Challenges== | ==Challenges== | ||
| Line 21: | Line 21: | ||
*Warlpiri is an Ergative-Absolutive language: "An ergative–absolutive language is a language in which the single argument ("subject") of an intransitive verb behaves like the object of a transitive verb, and differently from the agent of a transitive verb" (Wikipedia). | *Warlpiri is an Ergative-Absolutive language: "An ergative–absolutive language is a language in which the single argument ("subject") of an intransitive verb behaves like the object of a transitive verb, and differently from the agent of a transitive verb" (Wikipedia). | ||
*There's a small number of verb roots (about 100), which have tense markers, but there's a huge number of auxiliaries. | *There's a small number of verb roots (about 100), which have tense markers, but there's a huge number of auxiliaries. | ||
| − | *Auxiliaries (along with the verb suffix) identify tense, they identify the relationship between the main and dependent clauses, and they identify the person and number of the subject and object. | + | *Auxiliaries (along with the verb suffix) identify tense, they identify the relationship between the main and dependent clauses, and they identify the person and number of the subject and object. |
| + | **I have a small lexicon of auxiliary roots, and then a lexicon each for tense suffixes, subject suffixes, and object suffixes. The object and subject suffixes are both optional. | ||
*There are 11 different combinations of person/number, because there's a singular/dual/plural distinction as well as an inclusive/exclusive distinction for 1st person dual/plural. | *There are 11 different combinations of person/number, because there's a singular/dual/plural distinction as well as an inclusive/exclusive distinction for 1st person dual/plural. | ||
**I have a tag for each one of these combinations. | **I have a tag for each one of these combinations. | ||
| − | + | *A challenge that I'm facing with the disambiguation is that word order isn't based on grammatical function, but instead on topic/focus of the conversation. | |
| + | |||
| + | ==Who to work with== | ||
| + | *No one else is working on a language from the Pama-Nyungan language family. | ||
| + | *Thinking of working with Martina and Lee who have languages in the Austronesian family. I think they have had similar affixes identifying person/number as well as some common sources of ambiguity (such as limited distinction between nouns and adjectives). | ||
| + | *If anybody has had similar structures I'd like to hear from them! | ||
Latest revision as of 03:04, 2 March 2017
Notes for midterm evaluation
General
- Warlpiri is an endangered indigenous language of Australia
- Spoken by about 3,000 people in Australia's Northern Territory
- part of the Pama-Nyungan language family
- Warlpiri is an agglutinative language
- means that grammatical information is encoded in affixes attached to word stems
- Warlpiri words have many suffixes, so sentences have relatively few words
- Warlpiri has 21 phonemes, and uses a Latin orthography
Evaluation
- Currently getting about 40% coverage of my corpus. I increased this number dramatically this week by adding punctuation, because in Warlpiri, the suffixes are often added with hyphens, so most words contain hyphens.
- Total number of tokens: 1416 and total number of analyses: 1574 (which is puts my measure of ambiguity at 1.1) which means that I have some ambiguity to sort through.
Challenges
- Understanding the grammar has been the hardest part for me. Luckily Warlpiri is a quite well-studied language, so I've been able to find a few good resources. A few things that have been difficult:
- Warlpiri is an Ergative-Absolutive language: "An ergative–absolutive language is a language in which the single argument ("subject") of an intransitive verb behaves like the object of a transitive verb, and differently from the agent of a transitive verb" (Wikipedia).
- There's a small number of verb roots (about 100), which have tense markers, but there's a huge number of auxiliaries.
- Auxiliaries (along with the verb suffix) identify tense, they identify the relationship between the main and dependent clauses, and they identify the person and number of the subject and object.
- I have a small lexicon of auxiliary roots, and then a lexicon each for tense suffixes, subject suffixes, and object suffixes. The object and subject suffixes are both optional.
- There are 11 different combinations of person/number, because there's a singular/dual/plural distinction as well as an inclusive/exclusive distinction for 1st person dual/plural.
- I have a tag for each one of these combinations.
- A challenge that I'm facing with the disambiguation is that word order isn't based on grammatical function, but instead on topic/focus of the conversation.
Who to work with
- No one else is working on a language from the Pama-Nyungan language family.
- Thinking of working with Martina and Lee who have languages in the Austronesian family. I think they have had similar affixes identifying person/number as well as some common sources of ambiguity (such as limited distinction between nouns and adjectives).
- If anybody has had similar structures I'd like to hear from them!
