Word Frequency and the Generation Effect

Only available on StudyMode
  • Download(s) : 184
  • Published : April 5, 2012
Open Document
Text Preview


This report aimed to investigate the generation effect occurs for low frequency words. The experiment used a sample of 117 second year Research Method students from Birkbeck Univerity in within and between subject design. There were two independent variables, read and generate items and two dependant variables, low and high frequency. This data was analyzed with related sample t test to examine whether the generation effect occurs for low frequency words and independent sample t test to investigate whether there is a difference between generation effect with low and high frequency words. The results show that there is significant difference between generate and read condition for low frequency words and that the difference scores were not significantly higher for high frequency words than for low frequency words. These findings are discussed in terms of two theories of generation effect, namely the lexical activation hypothesis and the linkage associative hypothesis.


It is generally accepted that ‘learning by doing’ is more beneficial than passive reception of the same information i.e. reading. It is also believed that distinction of the words through bolding, highlighting or using custom fonts, facilitate perception and remembering of the transmitted content. The process called generation effect has been of interest of many scientists who examined memory performance and learning processes.

One of the first experimenters that tried to establish the difference between self-generated word and those externally presented were Slameka and Graf(1978). They used a variety of experimental manipulations i.e. rule of opposite (hot-cold), cued and uncued stimulus or within and between design. They found out that generation effect is an effect of practice in generation condition and that responses are better remembered than the stimuli. They also argued that the interaction between learning condition and rules make it difficult to determine the role of semantic processing in generating.

Later studies focused on meaningfulness of generated sentences(Graf, 1980) and another theory emerged, namely the lexical activation hypothesis proposed by McElroy and Slameka (1982). According to their theory, semantic processes are necessary for generation effect to occur. Therefore, it should not occur for meaningless word – non words. They run a number of experiments using non words under two experimental conditions: read vs. generate. At first they found out that there is no generation effects if the material is not in semantic memory. Therefore, read condition had better performance than generate condition. Slameka and McElroy concluded their findings by stating that lack of lexical status of the stimulus results with the absence of the generation effect with non words.

Nairne, Pusen and Widner (1985) investigated the lexical activation hypothesis and proposed associative linkage hypothesis. Their first two experiments focused on examining whether a knowledge of lexical status could influence generation effect with non words and whether providing a definition cue for non words would have an impact on remembering generated items better than read items. Their findings show no difference in instructions given to subjects of both groups as well as no generation effect for either group was found. However, the results of second experiment show significant difference between defined and undefined non words. Their third experiment was to determine the role of word frequency for the generation effect, assuming that it depends on a number of associates. This time participants were asked to switch underlined letters and to write down the resulting word or simply copy the word if not underlined. Paradigm used in this experiment was based on non words and low, medium and high frequency words. After a one minute distractor task, participants were given a...
tracking img