Documentation for the Grammar Matrix Customization Direct-Inverse Library

Introduction

This document explains how to fill out the Direct-inverse page of the Grammar Matrix Customization questionnaire and presents background information on the direct-inverse 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.

Citing the Direct-inverse Library

The standard reference for the Direct-inverse Library and its implementations is Drellishak 2009. The full reference and .bib entry can be found here.

Options

The direct-inverse library allows users to define the scale used to determine direct-inverse marking (if applicable). Please note that this page is optional. Leaving it blank will result in a grammar without direct-inverse marking.

In some languages, referred to as direct-inverse languages, marking of verbal arguments depends on grammatical scale (sometimes also referred to as hierarchy) that ranks agent and patient in terms of how natural they are in the role of agent. If in a sentence the agent ranks higher than the patient, the clause is considered to be direct. And, vice versa, if the patient ranks higher than the agent, the clause is considered to be inverse. Depending on the ranking, appropriate direct-inverse marking occurs through different verb forms, different cases on argument NPs, or both. The scales (hierarchies) vary for different languages, as well as does the corresponding marking.

The Grammar Matrix Customization system covers only one type of inverses, i.e. pronominal or morphological inverses. The other type of inverses, i.e. word order inverses, is not yet covered by the Grammar Matrix customization system (for more information please refer to the Upcoming Work section.

If your language is not a direct-inverse language, please skip this section entirely and move on to the next page Tense, Aspect, and Mood.

If your language is a direct-inverse language, you can add the features that define the direct-inverse scale on the Direct-inverse page of the Grammar Matrix Customization questionnaire. NOTE: Even if your language is not described as a direct-inverse language, but also uses a grammatical scale to determine marking on other parts of speech, such as NPs, you can still benefit from modeling it as a direct-inverse language (please see Motivation section below for an example of such language, Fore).

An example of a direct-inverse language is Cree (Algonquian language of North America), in which the direct-inverse scale is sensitive to person, with the following ranking of person:

As shown above (and described in greater detail in Drellishak 2009, Chapter 4), in Cree 2nd person ranks higher in terms of fulfilling the agent role than the 1st person, which in its turn ranks higher than the 3rd proximate person, and so on.

As instructed in the questionnaire, please make scale entries in order from the highest to the lowest ranking. Once the scale is defined, it will be available for use on the Morphology page to create the appropriate lexical rules. To model the direct-inverse pattern in Cree shown above you would add 4 scale entries, starting from the highest ranked value (2nd person) and finishing with the lowest ranked value (3rd person obviative).

Next you will need to define the behavior of the verb when patient and agent have equal rankings on the grammatical scale, i.e. whether the verb takes direct form or some other form, e.g. reflexive. For Cree, for instance, a user would indicate that the main verb takes some other form when agent and patient have equal ranking:

For more information on the implementation of direct-inverse pattern in Cree please refer to Test Cases in Drellishak 2009, Section 4.4.

Motivation

The Direct-inverse library was added to provide support marking on verbs and verbal arguments that is sensitive to a grammatical scale present in direct-inverse languages.

Additionally, the Direct-inverse library can be used for languages that are usually not analyzed as direct-inverse, but have similar characteristics and can therefore be modeled using direct-inverse analysis of the Grammar Matrix Customization system. For example, in the Fore language of Papua Guinea, the ranking of agent and patient determines the presence/absence of marker not on verbs, but on agent NPs. This marker, i.e. an extra suffix -wama appears on the agent when it has a lower ranking than the patient. Although Fore is not analyzed in standard literature as a direct-inverse language (Scott 1978, Drellishak 2009), it can nevertheless benefit from a direct-inverse analysis, with appropriate modifications to which parts of speech are marked based on the status of the clause (direct vs. inverse). For more information on modeling Fore using direct-inverse analysis please refer to Drellishak 2009, Sections 4.2.3 and 4.4.2.

Analyses

Defining a grammatical scale on the Direct-inverse page of the Grammar Matrix Customization questionnaire will result in the creation of appropriate lexical rules for transitive verbs. Additionally, more values will be available on the Lexicon page for your use, i.e. user can choose whether transitive verbs follow the direct-inverse pattern or not. The user will also be able to specify whether the verb features on the Lexicon page correspond to higher- or lower-ranked arguments.

Below is the snippet of code from direct-inverse section in the choices file for Cree language:

section=direct-inverse
scale1_feat1_name=person
scale1_feat1_value=2nd
scale2_feat1_name=person
scale2_feat1_value=1st
scale3_feat1_name=person
scale3_feat1_value=3rd
scale3_feat2_name=proximity
scale3_feat2_value=proximate
scale4_feat1_name=person
scale4_feat1_value=3rd
scale4_feat2_name=proximity
scale4_feat2_value=obviative
scale-equal=other

For the analysis of direct-inverse languages, a direct-inverse scale was modeled using a type hierarchy and arranged into a right-branching tree. This scale and its constraints preventing spurious parses are described in greater detail in Drellishak 2009, Section 4.2.

Additionally, a list feature [SC-ARGS] was introduced to support some direct-inverse languages, in which instead of agreement between the verb and subject or object, agreement is described as agreement between the verb and the higher- or lower-ranked argument. This feature contains the list of arguments of the verb in order from highest-ranked to lowest-ranked. This means that if the verb is direct, the subject will occupy the first position and the object will occupy the second position on the list. And, vice versa, for an inverse verb, the object will occupy the first position and the subject will occupy the second position on the list.

The analyses implemented in the direct-inverse library are described in greater detail in Drellishak 2009, Section 4.2.

Upcoming Work

One of the current limitations of Grammar Matrix Customization system is lack of support of word order inverses. Adding support for this grammatical phenomenon, involving interaction of inverses with word order, would be one of the possible directions of broadening Grammar Matrix coverage of direct-inverse languages.

References

Drellishak, S. (2009). Widespread but Not Universal: Improving the Typological Coverage of the Grammar Matrix. PhD thesis, University of Washington.

Givón, T. 1994. The Pragmatics of De-transitive Voice: Functional and Typological Aspects of Inversion. In T. Givón (ed.), Voice and Inversion, pages 3–44, Amsterdam: Benjamins.

Scott, G. (1978). The Fore Language of Papua New Guinea. Canberra, Australia: Pacific Linguistics.

MatrixDoc/DirectInverse (last edited 2013-03-01 12:01:23 by AntskeFokkens)

(The DELPH-IN infrastructure is hosted at the University of Oslo)