Phonemic and Semantic Verbal Fluency Tasks: Normative Data for Elderly Brazilians

This study aims to investigate the infl uence of sociodemographic characteristics on the performance of older people on two tasks of verbal fl uency and provide normative data for a Brazilian population of healthy elderly individuals with different educational levels. The initial sample included 521 individuals aged from 60 years, participating in the Program Family Health Strategy. Participants who had scores suggestive of cognitive decline on the Mini-Mental State Examination, depressive symptoms in Geriatric Depression Scale and self-reported neurological or psychiatric disorders were excluded. The fi nal sample consisted of 218 participants in phonemic verbal fl uency task (letters F, A and S) and 265 participants for semantic verbal fl uency task (animals). The performance in both tests was associated with age and education, but not with sex. Still, the education variable was shown to have a greater impact on scores in phonemic and semantic tests than age in both forms of evocation. The results of this study suggest the importance of providing normative data for elderly Brazilians appropriate to age and education on verbal fl uency tasks.

The verbal fl uency tasks are frequently used to evaluate cognitive changes, characterize the performance profi les in healthy aging and in neurodegenerative diseases (Di Biasio et al., 2012;Martyr et al., 2012).Fast and easy to administer, these tasks can be managed individually or as being present in some assessment protocols (Dubois, Slachevsky, Litvan, & Pillon, 2000;Fillenbaum et al., 2008).In general, the patient must evoke the greatest possible number of words in accordance with a predetermined rule.There are three types of verbal fl uency tasks: free (no predetermined criterion to evoke words), phonemic (words must start with a certain letter) and semantic (words must correspond to a particular semantic fi eld).Although the purpose of the activities is the same (evoking words), patients may present different performances according to the modality used and the type of stimulus chosen (Beausoleil, Monetta, Leblanc, & Joanette, 2001), since there are less frequent words that begin with specifi c letter (letter B versus X), as well as more restricted semantic fi elds (food versus color).In addition, sociodemographic characteristics, such as age and education, have been reported as variables that infl uence performance in both modalities (Bolla, Gray, Resnick, Galante, & Kawas, 1998;Brucki & Rocha, 2004;Carnero, Lendínez, Maestre, & Zunzunegui, 1999;Kempler, Teng, Dick, Taussig, & Davis, 1998;Silva, Yassuda, Guimarães, & Florindo, 2011).Other authors also point to the contribution of gender (Kempler et al., 1998;Silva et al., 2011).
Some studies suggest that the performance in verbal fluency modalities is a result of different cognitive functions that are activated during the task.In phonemic fl uency were evaluated primarily production and language fl uency skills, while in semantic verbal fl uency observed semantic memory skills (Peña-Casanova et al., 2009).However, both forms are associated with executive functions (Schlindwein-Zanini, 2010), since they require self-regulation, working memory, processing speed and organization (Magila & Caramelli, 2001).Functional neuroimaging studies highlight an increased activation of the frontal cortex to the evocation of words with phonemic criteria, while the temporal cortex is related to the recall of words with semantic criteria.Moreover, in both modes there is greater involvement of the left hemisphere (Gourovitch et al., 2000).
During the aging process, many factors can infl uence the performance of cognitive skills.The verbal abilities (e.g., syntactic production), however, remain preserved with normal aging.Nevertheless, there is a perceived diffi culty with advancing age in activities that relate language and memory (Parente, Saboskinsk, Ferreira, & Nespoulous, 1999).Thus verbal fl uency tasks are used as identifi ers of early cognitive decline (Chaves et al., 2011;Silva et al., 2011).Verbal fl uency tests are considered primarily as screening tools able to identify early cases of brain deterioration.The verbal fl uency appears altered in multiple pathological processes such as Alzheimer's dementia or frontotemporal dementia, in the left or bilateral frontal lesions, and in psychiatric disorders such as schizophrenia and depression (Rodrigues, Yamashita, & Chiappetta, 2008).Therefore this study aims to investigate the infl uence of sociodemographic characteristics on the performance of older people on two tasks of verbal fl uency (phonemic and semantic) and provide normative data for a Brazilian population of healthy elderly individuals with different educational levels.

Participants
The initial sample included 521 older adults, aged from 60 years, participants of the Family Health Strategy Program (FHSP; Gomes et al., 2013) in Porto Alegre, RS, Brazil.The FHSP randomly selected elderly residents of East/Northeast and Parthenon/Lomba do Pinheiro MD (evaluated by neurologists, psychiatrists, psychologists and nutritional experts).We excluded participants who had scores suggestive of cognitive decline in Mini Mental State Examination (MMSE) according to the cutoff points for education, scores ≥ 7 at the Geriatric Depression Scale of 15 points (GDS-15), and self-reported diseases neurological or psychiatric disorders.In the sample for the FAS verbal fl uency, illiterate participants were not included.The fi nal sample consisted of 218 participants in phonemic verbal fl uency task and 265 participants for semantic verbal fl uency task.Other sociodemographic informations of the participants are shown in Table 1.

Procedures and Instruments
After signing the informed consent form (research protocol approved -10/04967), participants were evaluated in a session of approximately 90 minutes duration.The older adults completed the sociodemographic and health questionnaire, MMSE, GDS-15, and two tasks of verbal fl uency: phonemic modality and semantic modality.
In phonemic modality used the FAS verbal fl uency task (Benton & Hamsher, 1976) which consists in evoking words that begin with the letters "F", "A" and "S", for a total of three minutes, one minute for each letter.The fi nal score refers to the sum of correct words spoken in each letter (Strauss, Sherman, & Spreen, 2006).In semantic modality used the Animals verbal fl uency task, which refers to evoke the greatest possible number of animals within 60 seconds.The total score refers to the number of animals cited correctly (Strauss et al., 2006).
The evaluations were conducted by students and professionals in the health fi eld such as medicine, psychology and speech therapy, trained for administration tasks.Corrections of tests were performed by a professional who did not participate in the data collection stage.

Data Analysis
Data were analyzed descriptively and inferentially.We used a multiple linear regression analysis with stepwise variable selection to assess the participation of the variables age, education and gender on performance in both verbal fl uency tasks.The dependent variables presented a normal distribution in the Kolmogorov-Smirnov Test (p ≥ .05).To compare the performance between age groups and gender on verbal fl uency tasks, FAS and Animals, we used the Student t Test for independent samples.To compare the schooling groups in semantic verbal fl uency task we use the One Way ANOVA analysis with Scheffe post hoc test.This same procedure was used to analyze the performance of different age groups distributed across schooling groups.Results were considered signifi cant when p ≤ .05.The statistical package used was the Statistical Package for the Social Sciences (SPSS) version 17.

Results
Table 1 presents the mean age, education, MMSE and GDS-15 scores of participant samples of verbal fl uency tasks.Regarding the distribution of gender, 139 (64%) were women and 79 (36%) were men among the participants who answered the FAS task.In the Animals task distribution was 164 (62%) and 101 (38%), respectively.Table 2 shows stepwise multiple regression analysis.According to the model investigated, the variables age and education demonstrate a signifi cant contribution to performance in both verbal fl uency tasks, and the number of years studied got greater infl uence than the age variable.However, the gender variable showed no signifi cant infl uence.
Comparisons between age groups, education and gender are presented in Table 3.According to analyzes for the FAS task, the elderly group aged 60-69 had signifi cantly higher performance at age ≥ 70.The group with more years of schooling achieved better results than those participants who studied only between 1 and 4 years of formal educa-tion.No signifi cant statistical differences were observed in relation to gender.
In the Animals task, there were no signifi cant differences between age groups and gender.However, the illiterate participants demonstrated signifi cantly lower scores than participants who had between 1-4 (p ≤ .001)and 5-8 (p ≤ .001)years of formal education.
Finally, in Table 4 sets out the normative data according to age groups and distributed as years of formal education.According to the ANOVA analysis signifi cant differences are observed when comparing the groups on both tasks verbal fl uency.

Discussion
The objective of this study was to investigate the effect of varying age, years of education and gender on performance on two tasks of verbal fl uency (FAS and Animals) in Brazilian elderly population neurologically preserved.From these data, we intended to present normative data adjusted for variables that were infl uential in the scores of these tasks.
In this study, the performance in both tests was associated with age and education, but not with sex.Still, the education variable was shown to have a greater impact on scores in phonemic and semantic tests than age, suggesting that the number of years of education predicts better results, like the lower the age.These results are similar to other studies that did not fi nd association between gender and the number of words generated, but the infl uence of age and education (Amaral-Carvalho & Caramelli, 2012;Munro et al., 2012;Ryu et al., 2012;Snitz et al., 2009;Venegas & Mansur, 2011).
The infl uence of education in the maintenance of cognitive abilities has received increasing attention (Meijer, van Boxtel, Van Gerven, van Hooren, & Jolles, 2009;Santos et al., 2012;Strout & Howard, 2012).This is due to the fact that the number of years of study could be a factor of cognitive reserve (Stern, 2009(Stern, , 2012)).According to the hypothesis of cognitive reserve activities such as reading, writing, learning languages, physical activity and recreation, for example, help to decrease the decline of cognitive functions.Thus, this could explain how some patients with sequel brain would show good results (Beausoleil, Fortin, Le Blanc, & Joanette, 2003).
However, Zahodne et al. (2011), through a longitudinal study, found that number of years of education infl uence performance on cognitive tasks, especially verbal fl uency, but not related to the changes over the years in any cognitive domain (verbal processing speed, working memory, verbal fl uency and verbal episodic memory).That is, the years of study are related to performance, but not with cognitive decline.Thus, individuals with higher education levels continue to show better cognitive performance than people with the same age and with less education, but decline at similar rates.
Regarding age, there is controversy between the results that combine the best results in cognitive abilities.Studies show that age is not a predictor of performance on tests of verbal fl uency (Brucki & Rocha, 2004;Venegas & Mansur, 2011).In contrast, Amaral-Carvalho and Caramelli (2012) found an association between age and performance of verbal fl uency tasks, as well as the results of other studies (Ryu et al., 2012;Silva et al., 2011;Snitz et al., 2009) corroborate the idea that the greater the age of the person, the smaller the words production of animals.
The difference in the effect of age on verbal fl uency tasks may have occurred because the categorical and semantics modalities activate different brain regions (Ryu et al., 2012).Robinson, Shallice, Bozzali and Cipolotti (2012) investigated how the frontal cortex behaves in different fl uency tasks (verbal, drawings, gestures, for example) with neurological patients (tumor and stroke).In verbal fl uency tasks observed that both types of verbal fl uency activates the frontal lobe, although the phonemic task was more specifi c in differentiating patients with frontal lesions of posterior lesions.This may have occurred because of the phonological modality requiring the creation of search strategies of words from a letter, making this task unusual and less automated search, resulting in increased activation of the frontal lobe (Rodrigues et al., 2008).
The results of this study demonstrate its importance in that it presents normative data on two tasks widely used in clinical practice and research to the cognitive assessment of older adults.Moreover, the performance standards that have been presented are appropriate for comparing clinical populations, since they were not included elderly with cognitive decline and depressive symptoms in the sample.A limitation of this study can highlight the lack of seniors with more years of schooling and the division of the sample into smaller age ranges.

Table 1
Description of Sociodemographic and Clinical Data of the Participants Note.NR = Not respond.

Table 3
Comparison of Performance on Verbal Fluency Tasks according to Age, Education and Gender Groups Note.Bold indicates signifi cant values.

Table 4
Normative Data according to Age Group and Distributed by Schooling