LONG-TERM MEMORY (LTM)
A store for permanent information, including world knowledge, the lexicon and general linguistic competence. In many accounts, LTM is distinguished from a sensory memory store of very brief duration, and from a limited-capacity working memory (WM) which holds currently relevant information and handles cognitive operations. LTM supplies information to WM when it is required and receives information from WM that is destined for long-term storage.
An item of information (e.g. a phone number or a name that we want to remember) can be consolidated and transferred from WM to LTM by rehearsal– by repeating it silently in our minds. Similarly, the more often we retrieve a particular item of information from LTM, the easier it becomes to access it and the less likely it is to be lost. Information that is rarely retrieved may decay, as in language attrition. Some accounts suggest that this is due to the loss of retrieval cues linked to the information sought.
LTM would appear to involve multiple memory systems, each with different functions. A distinction is made between two particular types of knowledge: declarative knowledge (knowledge that) and procedural knowledge (knowledge how). The first constitutes the facts we know about the world, and the events we recall; the second enables us to perform activities, many of which are automatic. Declarative knowledge is usually explicit and capable of being expressed verbally; it includes the kinds of grammar rule that a linguist might formulate. By contrast, procedural knowledge is implicit; it includes the ability to process language without necessarily being able to put into words the rules that are being applied.
In a classic account of how expertise is acquired, information is received into LTM in declarative form and gradually becomes proceduralised as WM makes more and more use of it. A novice first draws on declarative knowledge in the form of a series of steps towhich conscious attention (control) has to be given. In time, some of these steps become combined (composed), and the process becomes more and more automatic until it comes to form procedural knowledge.
Two types of declarative memory are generally recognised:
episodic memory stores events; it is specific in terms of time and place;
semantic memory stores generalised world knowledge.
The second may develop from the first. Imagine that a child stores in episodic memorya set of encounters with real-world entities that adults label DOG. Fromthese experiences, it can extrapolate a set of common features (or possibly a prototype); it thus forms a category in semantic memory which serves to identify the whole class of dogs. An alternative, exemplar-based view would minimise the role of semantic memoryand suggest that we identify examples of a category like DOG by relating them to many previous encounters with entities that have received this label, all of them stored episodically as individual events.
Semantic memory in LTM is sometimes represented as schematic in form. A schema is a set of interrelated features associated with an entity or concept. For example, the schema for PENGUIN might include: black and white– Antarctic– ice floe– fish– paperback publisher. Schematic information strongly influences the way in which we process incoming information, and is sometimes critical to the understanding of a text.
The ease with which a memory is retrieved from LTM is determined by how strongly encoded it is and by how precise are the available cues. Effective remembering may depend upon activating the same cues at retrieval as were originally encoded with the memory (the encoding specificity hypothesis). When subjects are asked to memorise the second words of some two-word compounds (e.g. STRAWBERRY JAM), the first word (STRAWBERRY) provides a powerful cue in later recall. However, the same does not occur if a different cue such as TRAFFIC is used.
See also: Memory, Schema theory, Working memory
Further reading: Cohen et al. (1993); Kellogg (1995: Chaps 4–6)
