Truly Smart Agents
projects > boas
boas

Boas is the knowledge acquisition component of Expedition. It guides source-language informants -- who need not be linguistic experts -- through questions about the ecology (letters, symbols, punctuation, etc.), inflectional morphology, derivational morphology and syntax of their language. It also sets them the task of building a complete bilingual closed-class lexicon and a large bilingual open-class lexicon.

Knowledge elicitation in Boas is largely driven by an inventory of parameters and values that are traditionally employed to describe natural language: e.g., case (nominative, genitive, dative, etc.), tense (present, past, future, timeless), phrase type (noun phrase, adjective phrase, etc.).

Examples of parameters, values and realizations that play a role in the Boas knowledge-elicitation process are shown in the table below. The first three rows illustrates inflection, the fourth and fifth, closed-class meanings, the sixth and seventh, ecology and the eighth and ninth, syntax.

ParameterValuesMeans of Realization
Case Relations nominative, accusative, dative, instrumental, abessive, etc. flective morphology, agglutinating morphology, isolating morphology, prepositions, postpositions, etc.
Number singular, plural, dual, trial, paucal flective morphology, agglutinating morphology, isolating morphology, particles, etc.
Tense present, past, future, timeless flective morphology, agglutinating morphology, isolating morphology, etc.
Possession +/- case-marking, closed-class affix, word or phrase, word order, etc.
Spatial Relations above, below, through, etc. word, phrase, preposition or postposition, case-marking
Expression of Numbers integers, decimals, percentages, fractions, etc. numerals in L, digits, punctuation marks (commas, periods, percent signs, etc.) or a lack thereof in various places
Sentence Boundary declarative, interrogative, imperative, etc. period, question mark(s), exclamation point(s), ellipsis, etc.
Grammatical Rolesubjectness, direct-objectness, indirect-objectness, etc. case-marking, word order, particles, etc.
Agreement (for pairs of elements) +/- person, +/- number, +/- case, etc. flective, agglutinating or isolating inflectional markers
Sample parameters, values and means of their realization.

Due to the extensive pedagogical support resident in Boas and its tightly organized method of elictation, Boas has been called a linguist in the box -- a tool providing practically the same level of assistance as would a human linguistic consultant. The knowledge-elicitation module was named Boas in honor of Franz Boas, a famous field linguist and pioneer of descriptive linguistics in the United States.

Boas was developed by a research team at NMSU CRL in 1997-2001. It is not currently available as a distributable system but we are seeking opportunities to reimplement it so that it could be.

relevant publications:

[1] McShane, M., Nirenburg, S., and Zacharski, R. Mood and modality: Out of theory and into the fray. 2004. Natural Language Engineering, 19(1): 57-89.

[2] McShane, M. and Nirenburg, S. Parameterizing the Space of Lexical and Grammatical Meaning Across Languages. 2004. Machine Translation. 18(2) : 129-165.

[3] McShane, M. An Innovative, Practical Approach to Polish Inflection. 2003. Lincom Europa.

[4] McShane, M. Applying tools and techniques of natural language processing to the creation of resources for less commonly taught languages. 2003. IALLT Journal of Language Learning Technologies 35 (1): 25-46. 

[5] McShane, M. and Nirenburg, S. Blasting open a choice space: learning inflectional morphology for NLP. 2003. Computational Intelligence 19(2): 111-135. 

[6] McShane, M. Redefining paradigm for (computer-aided) language instruction. 2003.Foreign Language Annals, 36 (2): 198-207. 

[7] McShane, M., Nirenburg, S., Cowie, J., and Zacharski, R. Embedding knowledge elicitation and MT systems within a single architecture. 2002. Machine Translation 17(4):271-305.

[8] Oflazer, K., Nirenburg, S., and McShane, M. Bootstrapping morphological analyzers by combining human elicitation and machine learning. 2001. Computational Linguistics 27(1).

[9] McShane, M. and Zacharski, R. Modularity in knowledge elicitation and language processing. 2000. Proceedings of the Third Annual High Desert Linguistics Conference, University of New Mexico, Albuquerque, NM, April 7-9, 2000, pp. 93-104.

[10] McShane, M., Helmreich, S., Nirenburg, S., and Raskin, V. Slavic as testing grounds for a linguistic knowledge elicitation system. 2000. In: Tracy Holloway King and Irina A. Sekerina, eds., Formal Approaches to Slavic Linguistics: The Philadelphia Meeting, 1999, pp. 280-295. Ann Arbor, Michigan: Michigan Slavic Publications.