Analogical Modeling: An Update

Analogical modeling is a supervised exemplar-based approach that has been widely applied to predict linguistic behavior. The paradigm has been well documented in the linguistics and cognition literature, but is less well known to the machine learning community. This paper sets out some of the basics of the approach, including a simplified example of the fundamental algorithm’s operation. It then surveys some of the recent analogical modeling language applications, and sketches how the computational system has been enhanced lately to offer users increased flexibility and processing power. Some comparisons and contrasts are drawn between analogical modeling and other language modeling and machine learning approaches. The paper concludes with a discussion of ongoing issues that still confront developers and users of the analogical modeling framework.