The Rey Auditory Verbal Learning Test: Normative data for the Arabic-speaking population and analysis of the differential influence of demographic variables

The Rey Auditory Verbal Learning Test (RAVLT) is the one of the most widely used neuropsychological tests of verbal memory. It has been translated into numerous languages including Spanish, German, Arabic, Hebrew, Czech, Portuguese, and English. The present study examined the hypothesis that the learning of word lists forms an algorithmic pattern across all cultures. To this end, a sample of 200 Arabic-speaking Omani adults between the ages of 18 and 50 years was collected. The resulting norms were then compared withexisting American and Brazilian samples. The study confirmed that the first trial on the RAVLT correlates with demographic variables, whereas the learning slope on subsequent trials is almost identical across all cultures. Based on the above finding, the slope of the verbal learning test is hypothesized to measure a psychophysiological process linked with the hippocampal formation and allows for the laying down of new memories. In contrast, the first trial of the test is amenable to more cultural, demographic, and environmental factors. Received 13 February 2012; received in revised form 30 May 2012; accepted 31 May 2012. Available online 29 June 2012. Amir Poreh and Alya Sultan, Neuropsychology Laboratory, Department of Psychology, Cleveland State University. Amir Poreh and Jennifer Levin, Department of Psychiatry, Case Western Reserve School of Medicine. Correspondence regarding this article should be directed to: Amir Poreh, Department of Psychology, Cleveland State University, 2121 Euclid Ave. CB 178, Cleveland, OH 44115. Phone: 216-6873718, Fax: 216 916-3704. E-mail: [email protected]. Introduction The Rey Auditory Verbal Learning Test (RAVLT) is one of the first commercially used measures of verbal memory. It was originally published in 1907 by Edouard Claparède and later translated to German, Italian, and Spanish (1919; see Boake, 2000). In 1958, Andre Rey, Claparède’s doctoral student, expanded the test and had patients repeatedly listen to and recall the same list of words five times. Following the presentation of a new list, Rey had his patient recall the previously presented words (Rey, 1958). That same year, Edith Taylor also added a 30-min delay trial and recognition trial (Taylor, 1959).Today, the RAVLT is considered the most widely used measure of verbal memory in the psychological literature (Mitrushina, Boone, Rzani, & D’Elia, 2005; Strauss, Sherman, & Spreen, 2006). To properly evaluate the performance of an examinee on this test, one usually calculates the deviation of the participant’s recall ineach session of the test from the mean recall in that session for normal participants with similar demographic characteristics. The means and standard deviations for different age groups are typically presented in a tabular form in numerous normative datasets (e.g., Lezak, Howieson, &Loring, 2004; Mitrushina et al., 2005; Strauss et al., 2006). Ideally, normative data should be based on tests of groups with a large number of participants matched on demographic variables that may affect performance on auditory learning tests such as age, gender, and intelligence quotient. Additionally, to be useful, the clinician must consider cultural differences as well as having adequate normative data with regard to the performance on such tests in the target population. The role of cultural differences in neuropsychological test performance is further complicated by the influence of education on one’s performance on such tests. For example, Manly, Schupf, Tang, & Stern (2005) noted that literacy level and not years of education appeared to be an important mediator of the interactions between biological and environmental factors.Thus, most recent normative studies, particularly those that used regression-based norms, use education level as a mediating factor when calculating within-subject test performance. A review of the literature shows that only a few studies attempted to translate neuropsychological measures into Arabic. Stanczak, Stanczak, &Awadalla (2001) translated and adopted the Expanded Trail Poreh, Sultan and Levin 58 Making Test to Arabic and examined the performance of a large sample of Sudanese subjects on this measure. The results showed that Americans exhibited markedly lower completion times than their Sudanese counterparts. Stanczaket al. (2001) attributed these differences to age, education, and sociocultural factors. Al-Ghatani, Obonsawin, & Al Moutaery (2009) and more recently Khalil (2010) examined the performance of healthy Arabic-speaking adults in Saudi Arabia on verbal fluency tests. Both studies used the Arabic version of the semantic (i.e., naming as many animals as possible) and phonemic (i.e., reciting as many words that start with a certain sound) versions of this test. In both studies, the average number of words generated in 60 secwas lower compared with English-speaking adults. To our knowledge, nostudy has yet attempted to develop normative data for the RAVLT for Arabicspeaking adults. When developing normative data for tests such as the RAVLT, distinguishing between the various components of the test is important. Many studies have shown that performance on the first presentation of the word list may be conceptualized as a measure of working memory or retention, and the slope that produces the five learning trials should be regarded as a measure of verbal learning.To clarify the strategies that underlie word list learning, Dunlosky & Salthouse (1996) decomposed the five-trial word list learning using lost and gained access indices. Lost access refers to the number of words recalled in a given trial that were not recalled during the immediately subsequent trial. Gained access reflects the ability to profit from the repeated presentation of the word list and refers to the number of new words recalled in a given trial that were not recalled in the immediately preceding trial. Dunlosky and Salthouse and other investigators showed that patients who suffer from dementia have low gained access and high lost access across trials. Poreh (2005) decomposed the learning trials using the best-fit linear logarithmic equation method. Using multiple North American normative samples (meta norms), Poreh demonstrated that the learning curve R(t) can be expressed by Equation 1: M(t) =R(t) = R1 + S ln (t ) [Equation 1] R1(trial 1) is the performance of a subject on the first trial.The coefficient S describes the mean rate of learning and henceforth is referred to as the logarithmic learning slope, learning slope, or slope. To test the above model, the normative data of North American English-speaking subjects were examined across the 58 published normative samples.The study showed an extremely robust curve fitting (mean r2 = .985, SD = .018).Additional analyses also showed that the immediate memory span on the first trial (R1) varies considerably with age and is affected by other background factors, whereas the mean learning slope (S) has a natural logarithmic curve (ln) and is practically independent of the mean age and other demographic factors. A weighted, linear, stepwise, multiple regression analysis of allofthe adult data yielded Equation 2: M(t) = R(t) = 4.015 0.0649 Age 0.0126 p.m.+0.30 Edu + 0.0224 IQ + S ln (t) [Equation 2] Poreh (2005) suggested that one may be able to use the values of R1 and S to evaluate patient performance instead of using their recall in each of the trials. Poreh also argued that the logarithmic learning curve reflects a fundamental psychophysiological process and would therefore emerge in every sample, regardless of the subject’s age, education, or gender. Support for Poreh’s (2005) distinction between the initial recall and learning of word lists was also separately reported by Jones et al. (2005). Using a single sample of North American older adults, they showed that the first trial on the RAVLT correlated with demographic variables. Additionally, they also demonstrated that the algorithmic curve fitting model provides the best fit model of verbal list learning. However, Jones et al. argued that the rate of learning was significantly associated with ethnicity, in addition to general cognitive status measured by the Mini-Mental State Examination. The first aim of the presentstudy was to publish norms for the RAVLT for Arabic-speaking adults. To this end, a large sample of Omani adults was administered the Arabic version of the RAVLT (A-RAVLT).The second goal of the study was to assess the hypothesis that the logarithmic learning curve on the RAVLT transcends ethnic groups and cultures. Much like their Englishspeaking counterparts, learning of the A-RAVLT list among Omani adults was predicted tofollow the same natural logarithmic curve pattern, whereas the first trial (R1) would be affected by demographic factors. Additionally, for cross-cultural comparison, the study examined curve fitting of the RAVLT learning curve using recently published RAVLT Brazilian normative data (Magalhaes & Hamdan, 2010).