Network Core Mechanisms of Exponence

Daland, Robert; Sims, Andrea D. and Pierrehumbert, Janet (2007)

Abstract

Against most of the literature, the authors argue that paradigmatic gaps can be learned from negative evidence: For high-frequency lexemes the learner estimates stochastically the expected frequency for a given form and detects a gap when the attested frequency is significantly lower. Low-frequency forms inherit defectivity by analogy to morphophonologically similar neighbors with high frequency. The approach is exemplified with Russian 1sg verb forms (cf. Halle 1973) for whikch it is argued that an account in terms of competing patterns (Albright 2005) is not viable. Finally a network simulation for the transmission of gaps along generation of speakers is presented which predicts stability of existing and creation of new gaps in the context of morphophonological similyrity to other gapped verbs.

Daland:2007
authorDaland, Robert and Sims, Andrea D. and Pierrehumbert, Janet
year2007
titleMuch ado about nothing: A social network model of Russian paradigmatic gaps
entrytypeinproceedings
editorZaenen, Annie and van den Bosch, Antal
booktitleProceedings of the 45th Annual Meeting of the Association for Computational Linguistics in Prague, Czech Republic, June 24th-29th, 2007
publisherPrague: Association for Computational Linguistics
pages 936-94

BibTeX

@inproceedings{Daland:2007,
    author = {Daland, Robert and  Sims, Andrea D.  and Pierrehumbert, Janet },
    year = {2007},
    title = {Much ado about nothing: A social network model of Russian paradigmatic gaps },
    booktitle = {Proceedings of the 45th Annual Meeting of the Association for Computational Linguistics in Prague, Czech Republic, June 24th-29th, 2007},
    editor = {Zaenen, Annie  and van den Bosch, Antal },
    publisher = {Prague: Association for Computational Linguistics},
    pages = { 936-94},
}
last changed: