time and their relationships with the integrity of white matter microstructures in four tracts related to semantic fluency. Results demonstrated that not only the total number of correct words for the animal and fruit fluency tasks but also the correct words generated at first 30 seconds in fruit fluency could significantly differentiate the MCI group from the CN and young adults groups. The finding that the total number of words generated of CN was fewer than young adults in both tasks is consistent with previous studies (Chan & Poon, 1999; Meinzer et al., 2009; Troyer, 2000; Troyer et al., 1997). We found that the MCI group produced significantly fewer total number of words in animal fluency than the CN group is consistent with previous studies based on aMCI (Joubert et al., 2010; Nutter-Upham et al., 2008;
Weakley et al., 2013) and possible and probable AD populations (Raoux et al., 2008).
The identical pattern of group differences for the total number of words generated in the animal fluency task was also found in the fruit fluency task which was in
accordance with the results of relative few studies investigating the fruit fluency task in MCI (Radanovic et al., 2009) and preclinical AD (Saxton et al., 2009).
With respect to the category fluency performances across the two time intervals, as expected, all groups generated more words at interval I than at interval II in both tasks which was consistent with the time-dependent retrieving processes (Fernaeus &
Almkvist, 1998). However, the MCI group generated fewer words during the first 30 seconds than CN in fruit fluency. The word productivity was thought to begin with a semi-automatic and rapid process in retrieving the words already available from semantic memory. During the second phase, word retrieving process becomes more effortful and needs to drive on more extensive search in semantic subcategories in order to find new words. Our data at the interval I in the fruit fluency task
implicated that the MCI group may have decreased efficiency in retrieving available or high-frequency words from the semantic memory. However, the MCI group (mean ratio=0.09, SD=0.06) was comparable to the CN group (mean ratio= 0.14, SD=0.07) at the interval II (p=0.08) in the fruit fluency task may partially due to relatively smaller sample size with reduced power for detecting the effect and the arrival at the asymptomatic level after 30s. To our knowledge, there was no other study investigated the performance of fruit fluency at the first 30 seconds in MCI.
For animal fluency, our result showed that the MCI group was comparable to the CN group at the interval I which is consistent with one study that investigated animal fluency under two time frames suggesting there was no difference at the first 30
seconds or the later 30 seconds in comparison of single domain aMCI, multiple domains aMCI, or non-amnestic MCI compared to controls (Weakley et al., 2013).
Although three groups consistently generated more words in animal fluency than fruit fluency at which is in accord with previous findings suggesting that it is easier to generate names for animals than for other categories (Juliana V. Baldo & Shimamura, 1998; Chan & Poon, 1999), such as occupation, fruit, or transportation. There were two possible explanations for the different results at the first interval for animal and fruit fluency tasks. First, the fruit fluency task may provide higher sensitivity than the animal fluency task in differentiating the MCI group from the control group, especially in higher-educated population. Secondly, the fruit fluency task may avoid generating a long string of exemplars (i.e., Chinese zodiac signs) that commonly happened in the animal fluency task which could potentially mask the actual sematic performances.
Previous research indicated that the category fluency task was associated with age, education, language, and culture backgrounds (Acevedo et al., 2000; Kempler et al., 1998). In this study, we already matched the education attainment and selected the animal and fruit categories that were known to be relatively stable and suitable for investigating cross-cultural (i.e., American vs. Chinese) or cross-age (i.e., young adults vs. old adults) differences (Yoon et al., 2004). In addition, Radanovic et al.
(2009) investigated the effect of education on the performances the animal and fruit fluency tasks in groups of healthy elderly controls, MCI, and AD groups. As a result, both tasks could differentiate AD from controls, but the fruit fluency task showed better sensitivity in differentiating MCI from controls. Moreover, the results suggested that the fruit fluency task was superior than the animal fluency task in differentiating higher-educated (more than eight years) participants, whereas the animal fluency task was superior than the fruit fluency task in differentiating less educated participants (four to eight years) (Radanovic et al., 2009). Consistent with the previous result, the average education years of participants in the present study was above thirteen years and the fruit fluency task showed bigger effect size in the total number of correct response (ƞ2=0.49) and response ratio at interval I (ƞ2=0.49) than the animal fluency task. In addition, Chiu et al. (1997) investigated the verbal fluency performances between normal aging and demented participants in Hong-Kong populations using three categories (i.e., animal, fruit, and vegetable), and the fruit fluency task was administrated for only 30 seconds, which was notable due to the concordance with the duration of interval I in our study. Their results showed that the performance in the fruit fluency task obtained the highest sensitivity (94.6%) among the other two categories (i.e., animal: 83.9%; vegetable: 78.2%) and the three categories scores combined (80.0%) in differentiating demented participants from the
normal aging group (Chiu et al., 1997). Secondly, previous study investigated the animal fluency performances in different ethnicities, including American, Chinese, Hispanic, and Vietnamese, and found that the ethnicity did not affect the total number of word generated in the animal fluency task (Kempler et al., 1998). For example, they found that Chinese participants produced words in relatively restricted set of animal names compared to American, but the two groups were not different in the total number of words generated. The authors explained that the difference in the variety of the animal names might be due to the exposure to a larger number of animals in American’s daily life by textbook, media, and zoos. However, a study indicated that the influence of Chinese zodiac signs on the animal fluency task was not only in Chinese but also in Korean cultures (Kim & Kim, 2012), and the results found that half of the top twenty highest production frequency animal names was the names of Chinse zodiac signs in which are the identities of twelve animals given to Chinese lunar years, including mouse, cow, tiger, rabbit, dragon, snake, horse, sheep, monkey, chicken, dog, and pig. In addition, evidence showed that animals within the same subcategory are closely related in semantic memory, which means that once an animal name was retrieved, it becomes easier to retrieve other animal names from the same subcategory (McClelland & Rogers, 2003) due to semantic proximities between items. Proximity could be estimated by the frequency of co-occurrence, sequential
dependency, correlation, or distance (Schvaneveldt et al., 1989). As for fruits, evidence showed that fruits have relatively few distinct features with weak semantic association (Tyler & Moss, 2001), and was supported by the findings that concepts of the fruit category was more sensitive than animal category in detecting sematic impairments in simplex encephalitis patients with different severity in anterior
temporal lesions (Bunn et al., 1998; Laiacona et al., 1997). As a result, we suspected that the fruit category may need more effort to retrieve a name for the decreased semantic proximities between different fruit exemplars compare to the animal category, therefore, the fruit category fluency task may be more sensitive to detect patients who are in early stage of the MCI-AD spectrum. Moreover, our data also showed that there were five young adults, seven CN, and ten MCI participants
generated twelve animal names of Chinese zodiac signs and mostly at interval I in the animal fluency task which may mask the semantic deficit in MCI. As a result, we could applied the first 30 seconds of the fruit fluency task in screening the MCI group due to the limited time in efficient clinical assessments, but if the time is abundant for more comprehensive evaluation, both animal and fruit fluency tasks may provide high sensitivity in detecting the difference between the MCI and the CN groups in clinical fields.