Chapter 2 The Theoretical Framework

This chapter will make a detailed introduction of and comparison between the representative models concerning the processing of regular and irregular forms, thus providing a theoretical framework for the present study. As mentioned in 1.1, several approaches address the issue of the processing of regular and irregular forms. Two of them have won greater popularity: the single-mechanism approach, especially the connectionist models (e.g., Rumelhart & McClelland, 1986; Plunkett & Juola, 1999; McClelland & Patterson, 2002), and the dual-mechanism approach, including the Words and Rules Theory (e.g., Pinker, 1991, 1998, 1999, 2006; Pinker & Ullman, 2002) and the Declarative/Procedural Model (e.g., Ullman, 2001a, 2001b, 2004). They disagree as to whether regular and irregular forms of a language are processed in one or two routes. The single-mechanism approach assumes that both regulars and irregulars are generated and parsed either only by a rules system or only by an associative memory system, whereas the dual-mechanism approach proposes that the regulars are generated and parsed by rules but the irregulars are arbitrary parings of sound and meaning, and therefore must be stored in and retrieved from memory.