LIFTing LEGO with RELISH: Lexicon Interchange FormaT in Use

Helen Aristar-DryInstitute for Language Information and Technology

Eastern Michigan U

Outline Background: The RELISH project The LEGO project

Project interface Project workflow

LIFT: Lexicon Interchange Format What it is Use in the LEGO project (“LL-LIFT”) Sample entry: comparison with LMF-

compliant XML output by LEXUS

The RELISH Project RELISH: Rendering Endangered Lexicons

Interoperable through Standards Harmonization Joint project: U of Frankfurt, MPI-Nijmegen,

LINGUIST List Goal: markup harmonization at two levels

Semantic Structural

Test cases: 6 lexicons fr. LEGO & Lexus

RELISH: MPI, ILIT and Lexicon Standards

LEGO & RELISH

LEGO TestLexicons

LEXUS Test Lexicons

- LL-LIFT- GOLD

- LMF-compliant XML (various)- ISOCats

RELISH

Interchange format: TEI? LIFT?

The LEGO Project 3-year project sponsored by the NSF Participants: ILIT (Linguist List) & U at Buffalo Goal: a “datanet” of interoperable lexicons. Interoperability based on: grammatical information mapped to GOLD structure mapped to a common schema (LL-LIFT) output in RDF

The LEGO Project Initial set of lexical resources: 17 EL lexicons prepared by LINGUIST List: Shoshone, Archi, Kayardild, Fulfulde, Mocovi, Biao

Mien, Potawatomi, Saliba, W. Pantar, W. Sissala, Wichi, …

3000+ wordlists prepared by U at Buffalo:Usher-Whitehouse lists, Loanword Typology Lists, Intercontinental Dictionary Series lists.

Extensible: In practice: to Lexicons in LIFT Possibly: RDF, TEI (or another official serialization

of LMF)

The LEGO Project Purposes Not intended to develop a lexicon creation or lexicon display tool

Intended to support multi-lexicon search and comparison

demonstrate the value of digital standards in linguistic research

LL

Descriptive XML

LL-LIFT (XML)

LEGO db

mapping to GOLD

LEGO Workflow

LEGO Interoperable

Lexicon

AccessLexicon - MS Word Excel Toolbox Filemaker

LEGO: lego.linguistlist.org

LEGO Browse Lexicons Page

LEGO: Lexicon ‘homepage’

LEGO: Browse Lexicon Entries

Demonstration of Interoperability: LEGO Search

LEGO: Search multiple lexicons by LIFT field (Definition, Example, Variant, etc.

LEGO: Search multiple lexicons by grammatical information (GOLD concept).

Extending LEGO: Upload Lexicon or Wordlist

Map to GOLD: Choose Lexicon to Edit

Mapping 1: View lexicon’s labels for grammatical concepts

Mapping 2: Click to view GOLD concepts, Click ‘Add’ to Map

Mapping 4: Lexicon label mapped to GOLD concept

Mapping 3: Access GOLD interface (if needed)

LIFT LIFT = Lexicon Interchange Format XML format for storing and exchange of

lexical information Developed by SIL International Designed to be easy to convert into and

out of MDF and Fieldworks formats Current version:

http://code.google.com/p/lift-standard/downloads/detail?name=lift_13.pdf

Programs that support LIFT

WeSay uses LIFT as its primary format. FieldWorks Language Explorer (FLEx)

 can import and export LIFT files. Lexique Pro can open LIFT documents

for viewing, printing, and making web pages. It can also save to LIFT format.

(fr. http://code.google.com/p/lift-standard/)

Utilities for LIFT

Solid can convert basic SFM (standard format markers, e.g. Toolbox format) to LIFT (see: http://lingtransoft.info/apps/solid

LiftTweaker Can selectively modify a LIFT file for targeting different audiences

Lexicons in LIFT LIFT chosen as upload format for LEGO

because of the large number of lexicons potentially available in LIFT About 50 published lexicons in Lexique Pro 180+ lexicons in Fieldworks Language

Explorer (FLEx) ? 300+ lexicons in Shoebox/Toolbox ?

With the owner’s permission, these could easily be integrated into the LEGO system

LIFT UML Diagram for Entry

Notes Grammatical Info is attached to Sense,

not to Entry or Form (differs from LMF) Variant is attached to Entry, not ‘sense’ –

can’t add a ‘sense’ to a variant Multiple senses and variants allowed Highly customizable: Field, Type, and

Range can be added to virtually any element (can be defined in the document header)

LL-LIFT Lack of constraint on use of Field

and Trait constituted a problem for LEGO

Developed ‘LL-LIFT’ a constrained form of LIFT which still validates against the

LIFT schema

LL-LIFT Major constraints

No Header Grammatical Information confined to a

single element Delimited within db field Parsed out during GOLD mapping

Minor entries, comparison forms, etc separate entries unified via ‘relation’ element

lexical-unit

dialects

sense

grammatical-info

definition

variant

relation

note

note

example

Structure of a LIFT Entry

paradigmatic traits

<entry id="d1e56244">

<trait name="original-id" value="2c9090a22632946601267b22f98e5098"/>

<lexical-unit>

<form lang="syv">

<text>дадагалзаар</text>

</form>

</lexical-unit>

<variant>

<form lang="syv">

<text>dadagalzaar</text>

</form>

</variant>

<sense>

<grammatical-info value="n."/>

<definition>

<form lang="eng">

<text>doubt</text>

</form>

</definition>

</sense>

</entry>

Tuva entry in LIFT

<LexicalEntry id="2c9090a22632946601267b22f98e5098">

<DataCategory type="rank">5155</DataCategory>

<DataCategory type="lexeme">дадагалзаар</DataCategory>

<DataCategory type="part of speech">n.</DataCategory>

<Form id="2c9090a22632946601267b22f9fd50a0" type="form">

<DataCategory type="image"/>

<DataCategory type="Audio">tvn_5155.mp3</DataCategory>

</Form>

<Sense id="2c9090a22632946601267b22f9fd50a6" type="sense">

<DataCategory type="gloss">doubt</DataCategory>

<DataCategory type="transcription">dadagalzaar</DataCategory>

</Sense>

<ListOfComponents/>

</LexicalEntry>

Tuva entry exported from LEXUS (LMF-compliant)

<entry id="d1e56244">

<trait name="original-id” value=“2…"/>

<lexical-unit>

<form lang="syv">

<text>дадагалзаар</text>

</form>

</lexical-unit>

<variant>

<form lang="syv">

<text>dadagalzaar</text>

</form>

</variant>

<sense>

<grammatical-info value="n."/>

<definition>

<form lang="eng">

<text>doubt</text>

</form>

</definition>

</sense>

</entry>LL-LIFT

<LexicalEntry id="2…">

<DatCat type="rank">5155</DatCat>

<DatCat type="lexeme">

дадагалзаар</DatCat>

<DatCat type="part of speech">n.

</ DatCat >

<Form id="2…" type="form">

< DatCat type="image"/>

< DatCat type="Audio">

tvn_5155.mp3</DatCat>

</Form>

<Sense id="2…" type="sense">

< DatCat

type="gloss">doubt</DatCat >

< DatCat type="transcription">

dadagalzaar</DatCat>

</Sense>

<ListOfComponents/>

</LexicalEntry>

LMF

Summary LL-LIFT = current XML schema used in LEGO Easy to transform (at least one) LMF-compliant XML output of LEXUS to LL-LIFT Transformation could be extended to TEI serialization of LMF. May require constraint