Syntactic systematicity arising from semantic predictions in a Hebbian-competitive network

• Published in: Connection science Vol. 13; no. 1; pp. 73 - 94
• Main Authors: Hadley, Robert F, Rotaru-Varga, Adam, Arnold, Dirk V, Cardei, Vlad C
• Format: Journal Article
• Language: English
• Published: London Taylor & Francis Group 01.03.2001; Taylor and Francis; Taylor & Francis Ltd
• Subjects: Applied sciences; Artificial intelligence; Competitive Learning; Computer science; control theory; systems; Connectionism; Connectionism. Neural networks; Exact sciences and technology; Language Acquisition; Language theory and syntactical analysis; Learning; Learning and adaptive systems; Neural networks; Semantic Features; Semantics; Systematicity; Theoretical computing; Learning; Competitive learning; Knowledge acquisition; Training analysis; Semantic feature; Systematicity; Connectionism; Language acquisition
• ISSN: 0954-0091; 1360-0494
• DOI: 10.1080/09540090110052996

A Hebbian-inspired, competitive network is presented which learns to predict the typical semantic features of denoting terms in simple and moderately complex sentences. In addition, the network learns to predict the appearance of syntactically key words, such as prepositions and relative pronouns. Importantly, as a by-product of the network's semantic training, a strong form of syntactic systematicity emerges. This systematicity is exhibited even at a novel, deeper level of clausal embedding. All network training is unsupervised with respect to error feedback. A novel variant of competitive learning and an unusual hierarchical architecture are presented. The relationship of this work to issues raised by Marcus and Phillips is explored.