-
Notifications
You must be signed in to change notification settings - Fork 4
MatrixDoc_Gender
This document presents background information on the gender library of the Grammar Matrix Customization System (Bender et al., 2002; Bender and Flickinger, 2005; Bender et al., 2010). General instructions on using the questionnaire can be found here.
[ This documentation is under construction. When it is more complete, this section should describe the effects of the various options provided in this library in terms of the behavior of the grammar. It is also a good place for tips on how to get the most from the library. ]
The gender library allows the user to specify the range of values for the feature GENDER that will be available elsewhere in the customization system (especially the Lexicon and Morphology pages) and in the resulting grammar.
On the gender page in Matrix Customization you are asked to describe the gender hierarchy in your language. Since, the number or genders and the complexity of gender hierarchy is different for every language, the number of genders and their supertypes that you can enter on Customization is unlimited. Such flexibility of Matrix customization mechanism allows user to cover languages with very complex or extensive gender hierarchies, like Swahili language.
Using dropdown menu, you will add genders and their supertypes as applicable to your language. If your gender does not express gender grammatically through agreement, you are asked to leave this section blank and move on to the next page. Below are examples of options chosen for two languages with different gender hierarchy: Russian language with a very straightforward one-level gender hierarchy and Tamil language with a more complex two-level gender hierarchy.
Russian (has 3 genders):
-
Gender Supertype Feminine gender Masculine gender Neuter gender
Tamil (has 4 genders, with rational gender further subdivided into feminine and masculine genders):
-
Gender Supertype Rational gender Feminine Rational Masculine Rational Neuter gender
For more information on how to use the questionnaire to define values for gender feature in your language, please refer to the Gender page in Matrix Customization.
For more information on the implementation of gender in Grammar Matrix please refer to the next section Analyses.
The analyses implemented in the gender library are described in Drellishak 2009, Section 5.3.
GENDER is a feature often involved in agreement phenomena. These are handled through the lexicon and morphology libraries. The analyses of agreement implemented in the Grammar Matrix are described in Drellishak 2009, Section 5.2.
In terms of filling out the questionnaire for Grammar Matrix, we are interested in gender only insofar as it is expressed grammatically, "Only in cases where we see covariation of an agreeing element are we dealing with grammatical gender“ (Drellishak 125).
Your grammar uses the feature GEND(ER), the values of which will be based on the values you have entered in the gender hierarchy. GENDER is a feature of the nominal INDEX (instead of the nominal HEAD). This allows for correct semantic representation of gender in your grammar (Drellishak 128).
The choices you make in the questionnaire are recorded in the choices file, which is a Unicode text file. Below is an example of how gender section in the Russian choices file would look:
section=gender
-
gender1_name=masc
- gender1_supertype1_name=gender
gender2_name=fem
- gender2_supertype1_name=gender
gender3_name=neut
- gender3_supertype1_name=gender
This will be reflected in the russian.tdl file:
png :+ [ PER person,
- NUM number,
GEND gender,
ANIMATION animation ].
;;; Gender
- gender := *top*.
masc := gender.
fem := gender.
neut := gender.
- [ This documentation is under construction. When it is more complete, this section should describe any modifications to or enhancements of this library that are either in progress or planned. ]
Gender implementation seems rather straightforward and the flexibility of Grammar Matrix customization mechanism already allows for an unlimited number of genders and their supertypes. The only possible direction of the upcoming work could be to allow the linguist to revise gender choices, if new information or a more complete linguistic analysis appears (Bender & Flickinger 2005, p. 2). (where would we want this flexibility? what was meant here?)
Bender, E., & Flickinger, D. 2005. Rapid prototyping of scalable grammars: Towards modularity in extensions to a language-independent core. In Proceedings of the 2nd International Joint Conference on Natural Language Processing IJCNLP-05 (Posters/Demos), Jeju Island, Korea.
Drellishak, Scott. 2009. Widespread but Not Universal: Improving the Typological Coverage of the Grammar Matrix. PhD thesis, University of Washington.
Home | Forum | Discussions | Events