The participants were recruited from community volunteers in Beijing. The eligible samples for inclusion were (1) 55–85 years old, (2) native Chinese speakers, and (3) the Mini-Mental State Examination (MMSE) ≥24 (22). Neurologists interviewed all the subjects. Any individuals with a history of psychiatric or central nervous system diseases, hearing loss, learning disability, and any other condition that was likely to have an influence on performance in the BNT were excluded.

The BNT-C used in this study consists of 30 items (Table 2) selected from the original 60 items without item adaption (20). The order of presentation followed the original sequence, and the design of the stimulus cards was identical to that of the original pictures (20).

The BNT-C was administered to all the participants by trained raters as described by Cheung RW who, together with his colleagues, developed the BNT-C test (20). All 30 cards bearing the line-drawing objects or animals were presented to the elders in a fixed order. Participants were instructed to name each object depicted on the cards. If the participant named one item correctly, one point was awarded and it was recorded among “scores of spontaneous naming (SN).” The examiner then proceeded to the next item. If the participant gave a wrong response or gave no response within 20 s, the participant's response was recorded in detail and a standard semantic cue was provided (e.g., “it is a plant” for “tree”). A semantic cue was designed for each item as in the original version of BNT. If the participant gave the correct answer, one point was awarded and it was recorded among “scores after semantic cue (SC)” (20). In the original BNT, if the participants failed the semantic cue, then a phonemic cueing was supplied. Given that Chinese is a logographic language and the names of most objects consist of one sound, BNT-C adopted a multiple-choice recognition task including the target response, name of an object similar to the target in function (e.g., “cow” for “camel”), and name of an object similar to the target in appearance (e.g., “mountain” for “camel”). The participant's response was recorded but no credit was given for choosing naming. The total score of BNT-C ranged from 0 to 30 (“SN” plus “SC”), the higher indicating the better naming ability.

The participants were instructed to rate each item of BNT-C for familiarity based on how usual or unusual the objects were in their experience, using a 5-point scale ranging from 1 (very unfamiliar) to 5 (very familiar). Familiarity was defined as “the degree to which you come in contact with or think about the object.”

Statistical analyses were performed with SPSS version 24.0 (SPSS Inc., Chicago, Ill., USA). Demographic and neuropsychological data were presented as mean ± SD or number and percentage. The mean data between two groups were analyzed using an independent-sample t-test or χ² test (chi-square). The internal consistency of BNT-C was assessed by Cronbach's alpha. Pearson's correlation coefficient was used to determine the correlation between naming accuracy and familiarity. A multiple linear regression analysis was used to explore the influence of age, gender, and education on BNT-C performance. For all tests, P < 0.05 was considered statistically significant.

The 30 items composing the BNT-C showed a high-internal reliability coefficient (α = 0.738). Every item in the BNT-C was positively correlated with the total score and contributed positively to Cronbach's α for the total score.

Participants' familiarities for each BNT-C item were recorded and compared with those from a study that rated the 60 BNT pictures on familiarity in 30 elder native English speakers in America (21). However, the study adopted a scale ranging from one (not at all familiar) to seven (very familiar) instead of one to five. To facilitate comparison, we multiplied their results by five-sevenths. The average familiarity of Chinese in our study was 4.33, but it was 4.94 in the American residents. Our elders rated not familiar (<4) in seven items (igloo, harp, pyramid, seahorse, dart, cactus, and trellis) and rated familiar or very familiar in the remaining 23 items. American elders rated all 30 items as familiar or very familiar. Further analysis into items with a familiarity difference >0.5 showed that Americans were much more familiar with igloos than Chinese (4.98 vs. 1.95). They were also more familiar with harp (5 vs. 3.07), pyramid (4.98 vs. 3.35), cactus (5 vs. 3.79), dart (4.86 vs. 3.71), seahorse (4.81 vs. 3.67), and trellis (4.93 vs. 3.83). The familiarities with rhinoceros, harmonica, tongs, protractor, and tripod were also 0.5 points higher in the Americans than in the Chinese residents. Abacus was the only item that the Chinese residents were more familiar with than the Americans (4.80 vs. 4.55) (Table 2).

Per item correct naming rates for the 30 BNT-C items were recorded and compared with data from American normal elders (23) and other Chinese normal elders outside the mainland [Taiwan (5) and Hong Kong (20)]. The correct naming rate difference between populations >20% was considered significant (5) (Table 2). Compared with the American residents, Beijing elders performed better in naming for compass, abacus, and protractor but performed worse in naming igloo, harp, dart, trellis, seahorse, cactus, and pyramid. The naming accuracy gap of igloo, harp, dart, protractor, compass, and abacus between the Chinese and the Americans was close to or >40%. Only 3.9% of the Beijing elders named igloo correctly compared with 98% of the American residents. In contrast, American residents got 55% and 46.7% correct naming rates for abacus and compass, respectively, lower than those of the Beijing residents (97.1 and 93.2%, respectively). Compared with the Hong Kong elders, the Beijing residents performed better in naming mushroom and protractor but worse in seahorse and dart. The Beijing residents named better in protractor and worse in seahorse, igloo, harp, and trellis than the Taiwan elders. Compared with the American elders, the Beijing, Taiwan, and Hong Kong elders were consistently good in naming compass and abacus but consistently worse in igloo, dart, and harp. All populations performed well in naming tree, pencil, scissors, flowers, racquet, hanger, camel, saw, snail, funnel, escalator, wheelchair, with a correct rate of >90%.

Pearson correlation analysis was used to explore the correlation between correct naming and familiarity. The Pearson correlation coefficient was 0.962 (P < 0.001), which indicated that naming accuracy was highly correlated with familiarity.

Multiple linear regression analysis was used to explore the effect of age, gender, and education on BNT-C performance (Table 3). No significant correlations between age or gender and BNT-C scores were found. The analysis identified that education strongly correlated with BNT-C performance. Therefore, the subjects were subgrouped into the following four educational categories: (1) ≤ 6 years (n = 32); (2) 7–9 years (n = 40); (3) 10–12 years (n = 42), and (4) ≥13 years (n = 47). According to the post hoc analysis, there were no significant differences between ≤ 6 years and 7–9 years of education and between 10–12 years and ≥13 years of education in scores for spontaneous naming (≤6 vs. 7–9 years: 20.29 ± 3.95 vs. 22.66 ± 3.31; 10–12 vs. ≥13 years: 25.00 ± 2.28 vs. 25.30 ± 2.37) and total scores after semantic cueing (≤6 vs. 7–9 years: 22.29 ± 3.58 vs. 23.94 ± 3.49; 10–12 vs. ≥13 years: 26.38 ± 1.83 vs. 26.65 ± 1.85). Accordingly, we subgrouped the participants into two education level groups (≤9 and >9 years) eventually.

The naming accuracy was significantly higher in the high education group (>9 years) than in the lower education group (≤9 years) in seahorse (77.00 vs. 42.90%, P < 0.001), dart (68.90 vs. 47.60%, P = 0.031), rhinoceros (95.10 vs. 78.60%, P = 0.010), harp (54.10 vs. 26.20%, P = 0.005), pyramid (86.90 vs. 47.60%, P < 0.001), compass (100.00 vs. 83.30%, P = 0.004), tripod (90.20 vs. 52.40%, P < 0.001), tongs (91.80 vs. 71.40%, P = 0.006), and protractor (85.20 vs. 57.10%, P = 0.001) (Table 4). Consequently, the high education group achieved better spontaneous naming scores (25.14 ± 2.33 vs. 21.79 ± 3.75, P < 0.001) and total scores (26.47 ± 1.88 vs. 23.36 ± 3.65, P < 0.001) than the lower education group (Table 1). We chose “mean-2^*SD” as the recommended cut-off value. For the individuals with formal education >9 years, the appropriate cutoff was 20 for spontaneous naming scores and 23 for total scores. For the individuals with formal education ≤ 9 years, the cutoff scores were 14 and 16, respectively.

BNT consists of 60 items for participants to name, including common things such as beds and trees, as well as uncommon objects such as pyramids and sphynx. People vary in their familiarities with each item due to different cultural backgrounds, life experiences, and education levels. However, familiarity data have been rarely evaluated and reported previously both in China and other countries. The current study found that the Chinese elders were not familiar with igloo, harp, pyramid, seahorse, dart, cactus, and trellis, which are not common in daily life or ordinary readings in China. Ferraro et al. rated the BNT pictures on familiarity in 30 elder native English speakers in America (21). As expected, the Chinese residents were less familiar with many items than the American residents. Most Chinese elders were very unfamiliar with igloo, which was very familiar to the Americans. In contrast, the Chinese residents were more familiar with abacus which was less familiar to the Americans. Many Chinese elders used the abacus for calculation in their early lives. The discrepancies in familiarities for BNT items truly reflected the cultural difference between China and America. Besides BNT-C, there were many adapted versions outside North America. Most modified versions replaced part test items according to local cultures (12, 17, 18). It seems appropriate to consider the possibility of replacing items that are not suitable for the Chinese individuals with more culturally representative items in the adapted Chinese BNT version. The development of new test items was outside the scope of this study, but future research should certainly address this consideration.

A previous study indicated that picture familiarity facilitates the processing of BNT word representations (24). It is expected that people likely name objects correctly when they are familiar with them. Consistently, we found a strong positive correlation between naming accuracy and item familiarity in the BNT-C performance. This correlation further explained the difference in per item correct naming rate between the Chinese and American residents. The American residents performed better in naming igloo, cactus, harp, and pyramid which they were more familiar with than the Chinese. In contrast, they perfomed significantly worse in naming abacus which they were least familiar with. The results further demonstrated how familiarity influenced naming accuracy.

BNT-C was developed by the Hong Kong scholars and was used in Taiwan and the mainland widely. Compared with people from Hong Kong or Taiwan, Beijing elders performed better in naming protractor but worse in seahorse, dart, igloo, and harp. This was probably because the residents in Hong Kong and Taiwan were more influenced by western culture than the Chinese mainland residents. Compared with the Americans, the Chinese people including Beijing, Taiwan, and Hong Kong consistently performed better in naming compass and abacus but consistently worse in igloo, dart, and harp. This consistency reflected the long-term differences between the Chinese culture and the American culture.

Hobson et al. (25) showed a reliable creation of an estimated 60-item BNT score from administrations of the 30-item BNT by multiplying the obtained score by two. By this method, the total BNT scores from our study and those from Hong Kong were multiplied by the two to allow comparisons across different studies. The calculated total scores (“scores of spontaneous naming” plus “scores after semantic cuing”) for elders in Beijing (estimated BNT score of 50) were lower than that in Hong Kong (20) (estimated BNT score of 53) and America (23) (estimated BNT score of 55). Combining the familiarity differences and the correlation between familiarity and naming accuracy, we could infer that cultural background influences participants' familiarities with BNT items, which affects the naming accuracy further and impact the total scores ultimately. However, it should be noted that the per-item familiarity and correct naming rate of BNT-C in this study and other studies in HK, TW, and the USA may have limited comparability given the varied age range and educational levels. Multicenter studies adopting the same inclusion criteria are required in future work for a more convincing comparison.

Demographic variables including age, gender, and education were repeatedly reported to impact the performance of cognition measures. Lower mean BNT scores with lower educational levels have been frequently found in the published works of literature (26–30). This study also demonstrated that education correlated significantly with BNT-C scores. Our findings were consistent with other research. Cheung et al. (20) performed the BNT-C in 77 normal adults in Hong Kong and found a positive association between education and naming scores (r = 0.342, P < 0.01). Chen et al. (5) applied the same BNT-C to 264 native Chinese people with normal cognition aged >60 years in Taiwan. They also found years of education were positively correlated with BNT-C score (r = 0.376, P < 0.01). Our research further showed that high-educated participants made fewer errors than low-educated subjects on dart, harp, pyramid, tripod, and protractor. The results were supported by the study exploring the influence of schooling on the performance of aphasia examination, which indicated that confrontation naming demands a greater degree of semantic knowledge, which is proved with increasing years of formal education (31). This may be because these objects are not common in daily life but are acquired gradually in study and reading.

The influence of gender and age on BNT performance remains controversial. Using the original 60-item BNT in normal elders in Middle Tennessee, Welch (30) found that age was significantly involved in confrontational naming ability. Moreover, there was also a gender bias that men scored significantly higher on 17 items than women. However, no gender effect was found for the Turkish version of the BNT (32) and the adapted BNT version for the Portuguese speakers (33). The Korean version of the BNT scores was slightly affected by age but rarely influenced by gender (10). For BNT-C, Cheung, who developed the BNT-C, found that gender and age have no effects on naming performance in Hongkong Chinese population (20). Chen et al. (5) did not identify the effect of either gender and age on BNT-C scores in the Taiwan residents. Consistent with previous research, we found no correlations between age or gender and BNT-C performance in this study. It likely indicated that gender and age affect the performance less on this 30-item BNT-C version.

This study has several limitations. First, our study mainly recruited the elders aged ≥55 years, so the data may not be generalized to the adults of all ages. Second, the major subjects were educated and only few were illiterate. This might underestimate the education effects on BNT-C performance. However, with the popularization of compulsory education, there were fewer illiterates now, and there will hardly be illiterates in the future. The results of this study will be suitable for future use. Third, the number of participants in each age-education cell was relatively small. Further research should expand the sample size. Fourth, the participants were recruited from communities in the urban areas of Beijing. In the future, urban and rural residents should be selected nationwide to make the results more representative. Another one that should be considered was that the American, Hong Kong, and Taiwan studies used for comparison were conducted 10 years or more before. The results may deviate from the real differences between the current population. However, the Chinese people rated less familiar and score on many items than the American decades later, indicating that cultural differences between China and America and their impact on BNT-C persist till now.