Memory-addressing Mechanisms and Lexical Access

This chapter discusses the proposal that the mental lexicon is only approximately content-addressable and that serial search mechanisms are inevitably involved in lexical access. The study of lexical access is important for two reasons. The lexical domain is well suited to an experimental study of this problem, because words form a well-structured and easily manipulated set of patterns. The central concept that integrates much of the theoretical and empirical work in this area is the concept of content-addressable memory. Content-addressability can be achieved in two ways, depending on the computing hardware available. The first involves building a totally different kind of memory, commonly referred to as an “associative memory.” Neural nets are a good example of this type of device. In this case, the “key” that unlocks memory is a partial description of the pattern that has to be retrieved, rather than its address. However, if the available memory is not associative, but is strictly location addressable, then content-addressability must be achieved through software rather than hardware. Such programming techniques are referred to as hash-coding. The essential part of hash-coding is a function that is designed to translate the input pattern into a unique address.