The Mechanism of Reverse Word Length Effect of Chinese in Working Memory

Abstract

One theoretically influential and much studied phenomenon related to working memory is the word-length effect:the finding that lists of short words are recalled at substantially higher rates than lists of long words (Baddeley,Thomson, Buchanan,1975). However,researchers have also found a reverse word-length effect for mixed word lists:isolated long words are recalled better than isolated short words within the same list (Hulme,Neath,Stuart,Shostak,Surprenant, Brown,2006). Current theoretical accounts of the word-length effect can be divided into two very different forms of explanation:primarily list-based or item-based. According to list-based explanations,the word-length effect is the product of some overall characteristic of the list. For example,the classic phonological loop hypothesis posits that there is a positive correlation between the rate of subvocal rehearsal and overt pronunciation time (Baddeley,1986). In contrast,item-based explanations hypothesize that short words are inherently easier to recall than long words (Neath Nairne,1995). For example,the SIMPLE model assumes that word-length effects are due to item complexity,and this complexity affects the relative discrimination of items in a list. Unfortunately,the reverse word length effect contradicts models that explain the word-length effect in terms of list-based accounts of rehearsal speed or as item-based effects. Modifications to current models or new explanations should be generated. In the present study,three experiments were conducted not only to integrate list-based and item-based explanations to account for the reverse word-length effect but also to propose new ideas for the mechanism. The study compared serial recall of word lists with varying lengths. Materials included three types of words:two-character,three-character and four-character Chinese words. The words were combined into four sets of word list,each of which consisted of five words:a pure short word list,a pure long word list,a mixed list containing four long words (background) and one short word (isolated word),and a mixed list containing four short words (background) and one long word (isolated word). Fifty-eight undergraduates (28 males,30 females) participated in the study. Experiment 1 found a word length effect in pure lists and serial position effect in all lists. Furthermore,short background words were still better recalled than the long background words and a reverse word length effect was observed in isolated words:Isolated long words were recalled better than isolated short words,and the isolated words were recalled more accurately than background words. Experiment 2 reduced the length difference of total pronunciation time for the two types of mixed lists. The reverse word length effect was eliminated and the rate of recall was still higher for isolated words than background words. These findings indicated that total pronunciation time had a role in recall performance. Experiment 3 used a technique of output delay. After the final item disappeared,a white crossband lasting 2 seconds was presented in the same location of a computer screen. An attenuated reverse word length effect was observed. However,there were no significant differences in recall performance for isolated,pure,or background words. These findings indicated that aside from phonological encoding,visual encoding,contributed to the memory process. Taken together,the findings showed that the reverse word-length effect could be explained only if list-based and item-based theories were integrated. Furthermore,results implicated a hypothesis of competitive-complementary parallel processing (CCPP) in relation to immediate serial order recall. That is,working memory may involve multiple,parallel,and different encodings that are both competitive and complementary in memory processing.