Download: pdf (63 pages; 698 Kb)

Abstract: The forms of words as they appear in text and speech are central to theories and models of lexical processing, yet current means of representing wordforms are lacking in certain key aspects. In the present study, a connectionist model termed the wordformer is presented that learns wordform representations through exposure to strings of stress-marked phonemes or letters. A small-scale simulation is reported to demonstrate the mechanisms and efficacy of the wordformer, and to show how it overcomes problems with other means of wordform representation. Two large-scale simulations are reported that learn phonological and orthographic wordform representations, respectively, for nearly 75,000 English wordforms. Model analyses show that processing generalizes much better to well-formed pseudowords compared with ill-formed (scrambled) pseudowords. An empirical test shows that wordformer performance is correlated with well-formedness ratings of orthographic pseudowords. It is discussed how the wordformer may be integrated into broader models of lexical processing.

Copyright Notice: The documents distributed here have been provided as a means to ensure timely dissemination of scholarly and technical work on a noncommercial basis. Copyright and all rights therein are maintained by the authors or by other copyright holders, notwithstanding that they have offered their works here electronically. It is understood that all persons copying this information will adhere to the terms and constraints invoked by each author's copyright. These works may not be reposted without the explicit permission of the copyright holder.