When language comprehension goes wrong for the right reasons: Good-enough, underspecified, or shallow language processing

Abstract

This paper contains an overview of language processing that can be described as “good enough”, “underspecified”, or “shallow”. The central idea is that a nontrivial proportion of misunderstanding, misinterpretation, and miscommunication can be attributed not to random error, but instead to processing preferences of the human language processing system. In other words, the very architecture of the language processor favours certain types of processing errors because in a majority of instances, this “fast and frugal”, less effortful processing is good enough to support communication. By way of historical background, connections are made between this relatively recent facet of psycholinguistic study, other recent language processing models, and related concepts in other areas of cognitive science. Finally, the nine papers included in this special issue are introduced as representative of novel explorations of good-enough, or underspecified, language processing.

Keywords:

• Good-enough language processing
• Underspecified language processing
• Shallow language processing
• Psycholinguistics
• Language comprehension

I thank Fernanda Ferreira for helpful feedback on this introduction, and Marc Brysbaert, Cassie Palmer, Peiyun Zhou, Nayoung Kim, and Erika Hussey for invaluable assistance in putting this special issue together.

Additional information

Funding

These efforts were funded in part by a National Science Foundation (NSF) CAREER Award [grant number BCS-0847533].