Parsing

Research on syntax and parsing played a central role in the cognitive revolution and continues to occupy a central position in the study of language. While linguistically driven theories of syntactic representation have not proven sufficient to predict and explain the full range of experimental outcomes, they have provided invaluable insights to language researchers. Greater success has been achieved by combining insights from linguistics and cognitive psychology. Such approaches began to emerge in the 1970s, with the advent of serial, modular, syntax-first approaches, such as the garden-path theory, that were computationally tractable and plausible given human resource limitations. The 'second wave' of syntactic parsing theories arose as an important component of connectionist/neural network-inspired research in the 1980s and 1990s. These accounts differed from first-generation accounts in that they viewed syntactic parsing as being one component of a massively interactive parallel-processing system. According to such accounts, the parser responds to a wide variety of lexical and contextual influences, and evidence for such effects can be found in experiments involving verb subcategory preferences, lexical-semantic properties of words in important syntactic positions, and discourse and visual context. Recent emerging theories have added new perspectives by focusing on multiple cues to meaning, redundancy at lexical-semantic and syntactic levels of analysis, the importance of anticipatory processes, and the importance of experience in adult comprehenders. Thus, parsing theory continues to evolve and we can expect this area of inquiry to remain lively for the foreseeable future. (C) 2010 John Wiley & Sons, Ltd. WIREs Cogn Sci 2011 2 353-364 DOI: 10.1002/wcs.112