The Primacy Effect in Free Recall of Lists of Common and Proper Names: A Study on Young, Elderly, and Alzheimer’s Disease Subjects

According to a widely recognized observation, elderly people experience an increased difficulty in recalling proper names. However, quite surprisingly, few systematic studies have been devoted to determining whether proper name recall performance does, in fact, decline with age. Experimental investigations confirm diary studies (see, for reviews and specific criticisms on diary studies, Crook and West, 1990 and Cohen, 1994) in demonstrating that younger subjects perform better than older subjects. Yet there is no data whether any decline in performance differs between proper and common names. The nature of such decline is also still obscure. One of the problems with diary and experimental studies is to make sure that the naming tasks used to compare common and proper names are of comparable difficulty. It is unclear, for instance, how this could be obtained with pictures of faces on the one side and pictures of objects on the other. The solution to this problem is sought in the present study by adopting a free recall test of supraspan lists of names. One very prominent and reliable characteristic of free recall of words is the serial position function: subjects are more likely to recall the early items (primacy effect) and the late items (recency effect) than the middle ones. According to a somewhat simplistic distinction, that however holds for the present purposes, the primacy effect is attributed to storage in long-term memory and is associated with semantic coding, while the recency effect is attributed to short-term memory and is based on phonological coding. The purpose of our experiments was twofold: to look for primacy effect differences according to the nominal class (common vs proper names) and to check any nominal class difference in the serial position curve related to aging or Alzheimer’s disease-related memory impairment. Methods and results. Ten young (mean age 22.7 years), 10 elderly (74.5 years) and 10 Alzheimer’s disease (66.5 years) subjects participated in two experiments. In the first experiment we administered 20 lists of words. Ten of these lists consisted of 12 common names of the same category (e.g., animals). The remaining 10 lists consisted of 12 proper names (e.g., John). Common names and proper names lists were matched for word frequency, word length, and phonological complexity in each serial position. Lists were also balanced across position so that each serial position had the same overall difficulty. The second experiment aimed to study the effect of release from proactive interference. For this purpose we administered twenty 12-item lists. In 10 of these ‘‘mixed’’ lists the first 6 items were proper names. Common names belonging to the same category followed. In the remaining 10 lists the opposite was true. For both experiments, lists were administered orally, at the rate of one word per second. Subjects were instructed to repeat in any order as many words as possible, immediately after the end of each list presentation. The ‘‘primacy effect’’ was calculated as the sum of words recalled from the first six serial positions. The ‘‘recency effect’’ was defined as the number of words recalled from the last three serial positions. The ‘‘first position effect’’ was also computed. ANOVA revealed that common names were better recalled than proper names (p , 0.0001). This effect was present in all experimental groups. The primacy effect (see Fig. 5) was significantly greater in young than in elderly (p 5 0.004) or in Alzheimer’s patients ( p 5 0.0006). A similar analysis for the first position effect got similar results. In the ‘‘mixed’’ list experiment, release from proactive interference was measured from the recall difference between the 7th and the 6th serial position. All the groups showed a greater release from proactive interference when common names were in the second half of the list (p 5 0.0009). Conclusion. Recalling proper names is more difficult throughout life as compared to recalling common names matched for frequency, length and phonological complexity, even if the actual task is exactly the same for both categories. We attribute this difficulty to the different semantic organization in long-term memory that the two categories might enjoy (proper names, indeed, are said to have no semantics). This hypothesis is supported by recent findings in aphasic patients (Semenza, 1995).