Documentation for the Grammar Matrix Customization Case Library

Introduction

This document explains how to fill out the Case page of the Grammar Matrix Customization questionnaire and presents background information on the Case 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 Case Library

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

Options

The case library allows the user to specify the range of case values used in the grammar (if any), and the general type of the language’s system for marking core cases. On the Lexicon page, each verb type can have an argument structure specified. The options available there depend partly on the answers to the questions on the Case page, but there is always the option of defining a verb class that uses a different case pattern than those provided on the basis of the selection for core case marking.

The Grammar Matrix customization system provides you with nine pre-defined case system options covering the most commonly used case systems, as well as one option allowing you to define additional cases. Following Dixon’s terminology (Dixon 1994), cases are discussed in terms of the roles of the arguments: A (agent of a transitive verb), O (patient or object of a transitive verb), and S (subject of an intransitive verb).

  • None
    Please select this option if your language does not use a case system morphosyntactically. Instead of expressing it morphosyntactically, such languages determine the roles of verbal arguments by word order, intonation, or pragmatically.

    Nominative-accusative

    Please select this option if your language uses the nominative-accusative (also referred to as accusative) case system. In such case systems S and A are marked with the same case, while O is marked with a different case. In the menu provided to you in Grammar Matrix Customization system, please specify the name of the case taken by S and A (e.g. nominative, subjective), and the name of the case taken by O (e.g. accusative, objective). An example of a language with nominative-accusative case system is German.

    Ergative-absolutive

    Please select this option if your language uses the ergative-absolutive (also referred to as ergative) case system. In such case systems S and O are marked with the same case, while A is marked with a different case. In the menu provided to you in Grammar Matrix Customization system, please specify the name of the case taken by A (e.g. ergative, relative, or narrative), and the name of the case taken by S and O (e.g. absolutive, nominative). An example of a language with ergative-absolutive case system is Australian language Dyirbal.

    Tripartite

    Please select this option if your language uses a tripartite case system. In such case systems all three roles, i.e. S, O, A, are marked with different cases. In the menu provided to you in Grammar Matrix Customization system, please specify the name of the case taken by S (e.g. nominative, subjective), the name of the case taken by A (e.g. ergative, agentive), and the name of the case taken by O (e.g. absolutive, patientive). An example of a language with tripartite case system is Wangkumara.

The next four options are the subtypes of the split ergativity case systems, which are neither nominative-accusative, nor ergative-absolutive. These case systems can be conditioned by the following factors: (1). semantic nature of the main verb, (2). semantic nature of the core NPs, (3). tense, aspect, or mood of the clause, and (4). grammatical status of the clause (Dixon 1994:70). Please note that the latter two (case conditioned on tense/aspect/mood and case conditioned on grammatical status of the clause) receive a similar analysis, described in more detail in Drellishak 2009, Section 3.2.

  • Split-S

    Please select this option if your language uses a split-S case system. In such case systems there are two types of intransitive verbs: verbs with A-like marking on their arguments and verbs with O-like markings on their single arguments. In the menu provided to you in the Grammar Matrix Customization system, please specify the name of the case taken by A (e.g. ergative, agentive), and the name of the case taken by O (e.g. absolutive, patientive). An example of a language with split-S case system is Mandan.

    Fluid-S

    Please select this option if your language uses a fluid-S case system. In such case systems, in addition to the two types of verbs used in the split-S languages, there are also intransitive verbs with A- or O-like markings on their single arguments, depending on pragmatics. For example, if the subject controls the action, then A-like marking is used on the argument. If the subject does not control the action, then O-like marking is used on the argument. In the menu provided to you in the Grammar Matrix Customization system, please specify the name of the case taken by A (e.g. ergative, agentive), and the name of the case taken by O (e.g. absolutive, patientive). An example of a language with fluid-S case system is the North Caucasian language Bats.

    Split conditioned on features of the noun phrase arguments

    Please select this option if your language uses a split-N case system, in which use of case marking is determined by the nature of nominal arguments. In such case systems some NPs (e.g. pronouns) have nominative-accusative marking pattern, while other NPs (e.g. common nouns) use the ergative-absolutive marking pattern. In the menu provided to you in the Grammar Matrix Customization system, please specify the name of the case taken by S and A (e.g. nominative, subjective), the name of the case taken by O (e.g. accusative, objective), the name of the case taken by A (e.g. ergative, relative, narrative), and the name of the case taken by S and O (e.g. absolutive, nominative). An example of a language with split conditioned on features of the noun phrase arguments case system is Dyirbal.

    Split conditioned on features of the verb

    Please select this option if your language uses a split-V case system, where case marking is determined by the nature of nominal arguments. You will be able to define these features later on the Lexicon page. In such case systems, case markings are determined by the tense, aspect, or mood of the clause. In the menu provided to you in Grammar Matrix Customization system, please specify the name of the case taken by S and A (e.g. nominative, subjective), the name of the case taken by O (e.g. accusative, objective), the name of the case taken by A (e.g. ergative, relative, narrative), and the name of the case taken by S and O (e.g. absolutive, nominative). An example of a language with split conditioned on features of the verb case system is Indo-Iranian language Gujarati, with the nominative-accusative case pattern in the imperfective aspect and the ergative-absolutive case pattern in the perfective aspect.

    Focus-case

    Please select this option if your language uses a focus-case system, in which an additional case exists (sometimes referred to as focus case), the grammatical role of which is determined by the morphology of the verb. An example of a language with focus-case system is Tagalog, in which presence of one argument with focus-marking is obligatory for every clause. In the menu provided to you in the Grammar Matrix Customization system, please specify the name of the focus case in your language, the name of the case taken by A (e.g. ergative, relative, narrative), and the name of the case taken by O (e.g. accusative, objective).

    Additional Cases

    Finally, many languages have more cases than those that cover standard A, S and O marking. You can add any number of cases in the Additional Cases section on the bottom of the Case page. This option is useful for implementation of quirky case, i.e. the case taken by obligatory verbal arguments beyond cases marking intransitive subjects, agents, and patients (Levin and Simpson 1981).

    To describe quirky case, define the appropriate range of case values on the Case page, including extra cases if necessary (in Additional Cases at the bottom of the page). Then on the Lexicon page, select an argument structure without any case presets. Constrain the CASE value of each argument through the add a feature iterator.

The values you assign to the CASE feature will be available to use later in the customization system, especially the Morphology and Lexicon pages. The answers you provide on this page will determine the values available on the Lexicon page for the CASE feature. CASE will also appear as one of the features of the lexical rule types on Morphology page.

Further information about the options on the Case page is provided in in Drellishak 2009, Section 3.3.

Motivation

While some language do not have a case system, other languages that do use cases rely on them heavily. The case library of the Grammar Matrix Customization system was developed to cover case systems that are most commonly used across languages, providing user with nine pre-defined options in the questionnaire. To provide the user with more flexibility in coverage of the less commonly used case systems or of the quirky case, the Grammar Matrix Customization system also allows user to define additional case values instead of just relying on the pre-defined analyses. For more information on the analyses of case systems please refer to the Analyses section below.

Analyses

After you define a case system for your language, your starter grammar will include the feature [ CASE case ], with possible values based on the choices you make on the Case customization page.

Below is a snippet of code related to case from choices file for a language with nominative-accusative case system:

section=case
case-marking=nom-acc
nom-acc-nom-case-name=nominative
nom-acc-acc-case-name=accusative

This can be compared to code from the same section of the choices file for a language that does not use a case system:

section=case
case-marking=none

If no further specifications for case is defined on the Lexicon or Morphology pages of the Grammar Matrix customization system, the case hierarchy will be flat, i.e. all cases will be direct subtypes of the feature-value case.

case := avm.
nom := case.
acc := case.
gen := case.
dat := case.
abl := case.

A more elaborate hierarchy can be formed by indicating common underspecifications for lexical items or morphemes. For instance, the Latin word tempus can either be nominative or accusative, but not genitive, dative or ablative. If you define tempus in the lexicon with nom and acc as case values, the hierarchy will included an underspecified form for case that is either nominative or accusative:

case := avm.
nom-or-acc := case.
nom := nom-or-acc.
acc := nom-or-acc.
gen := case.
dat := case.
abl := case.

Following Pollard and Sag (1994), CASE is a feature of the syntactic HEAD. The cases of both arguments (subject and object) are specified on the verb lexical types. Grammar Matrix provides the following case-marking strategies in the Lexicon section: (1). marking morphologically on nouns, determiners, or (2). marking of whole NPs via case-marking adpositions, or (3). both (mixed marking).

  • (1). Morphological marking is implemented through lexical rules. These rules take a lexical item as an input, apply spelling change, constrain CASE feature to have a particular value. For languages with morphological case-marking strategy, nominal heads have a [CASE] feature that is specified by lexical rules for the appropriate case inflection or directly on a lexical type. Verbs are constrained to take arguments that are noun phrases.

    (2). Case-marking adpositions are implemented through lexical items taking nominal complements and specifying a particular case marking on this complement, as well as on the adpositional phrase. In this strategy verbs take adpositional phrases as their arguments, instead of noun phrases. [CASE] feature is specified on both nominal and adpositional heads. Case-marking adpositions are constrained to take nominal complements as their arguments and specify an appropriate case value on both the complement and the resulting adpositional phrase. Verbs are constrained to have adpositional phrases as their arguments, thus preventing noun phrases appearing as their complement.

    (3). Mixed marking is more complex and is reserved for languages that have both morphological and adpositional case-marking, e.g. Tagalog. As in the case-marking adpositional analysis, the [CASE] feature is specified both on nominal and adpositional phrases. Verbs can take different arguments, i.e. nouns, noun phrases or adpositional phrases, depending on the language. For more information on implementing the mixed case-marking strategy and its complications please refer to Drellishak 2009, Section 3.2.1.1.

For the detailed analyses of each of the case-marking systems covered by Grammar Matrix Customization system please refer to Drellishak 2009, Section 3.2.

Upcoming Work

Although Case library of Grammar Matrix Customization system is already quite flexible, providing user with nine pre-defined case systems and allowing user to create additional cases, it could be extended for a broader coverage of case and its interactions with other grammatical phenomena.

One of the limitations of the Grammar Matrix is its lack of coverage of the more complex argument-marking cases, e.g. some fine interactions between case-marking and verb forms. Adding support for these interactions between case and verb forms (or between case and other parts of grammar) would provide a broader coverage of the grammatical phenomena described in these sections, as well as make the interaction between these sections of Grammar Matrix Customization system more dynamic.

Another current limitation of Grammar Matrix customization system is lack of coverage of the syntactic ergativity. Although syntactic ergativity is an inter-clausal phenomenon and absence of its coverage does not yet affect grammars created by Grammar Matrix Customization system, it would be beneficial to implement it in the future. Providing support for syntactic ergativity would also aid in the enhancements to the sections on relative clauses, control, and binding, which can interact with syntactic ergativity. This could be achieved through the integration of the HPSG analysis of syntactic ergativity that has already been developed by Manning and Sag (1995).

References

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

  • bibtex:

    @phdthesis{Drellishak:09,
    author = {Scott Drellishak},
    year = {2009},
    title = {Widespread but Not Universal: Improving the Typological Coverage of the {G}rammar {M}atrix},
    school = {University of Washington}
    }

Dixon, R. M. W. (1994). Ergativity. Cambridge: Cambridge University Press.

  • bibtex:

    @book{Dixon:1994,
    author = {R. M. W. Dixon},
    year = {1994},
    title = {Ergativity},
    publisher = {Cambridge: Cambridge University Press}
    }

Levin, L. S., & Simpson, J. (1981). Quirky Case and Lexical Representations of Icelandic Verbs. Chicago Linguistic Society 17, 185–196.

  • bibtex:

    @book{levin:1981,
    author = {Lori S. Levin and Jane Simpson},
    year = {1981},
    title = {Quirky Case and Lexical Representations of Icelandic Verbs},
    conference = {In Proceedings of the Chicago Linguistic Society 17},
    pages = {185–196},
    }

Manning, C. D., & Sag, I. (1995). Dissociations between Argument Structure and Grammatical Relations. In Lexical and Constructional Aspects of Linguistic Explanation, pages 63–78, CSLI.

  • bibtex:

    @incollection{manning:1995,
    author = {Christopher Manning and Ivan Sag},
    year = {1995},
    title = {Dissociations between Argument Structure and Grammatical Relations},
    booktitle = {Lexical and Constructional Aspects of Linguistic Explanation},
    pages = {63-78},
    publisher = {CSLI Publications}
    }

Last update: 2013-10-01 by EmilyBender [edit]