Relationship between the Brazilian version of the Montreal-Toulouse language assessment battery and education, age and reading and writing characteristics. A cross-sectional study

ABSTRACT CONTEXT AND OBJECTIVE: There is growing concern about understanding how sociodemographic variables may interfere with cognitive functioning, especially with regard to language. This study aimed to investigate the relationship between performance in the Brazilian version of the Montreal-Toulouse language assessment battery (MTL-BR) and education, age and frequency of reading and writing habits (FRWH). DESIGN AND SETTING: Cross-sectional study conducted in university and work environments in Rio Grande do Sul, Brazil. METHOD: The MTL-BR was administered to a group of 233 healthy adults, aged 19 to 75 years (mean = 45.04, standard deviation, SD = 15.47), with at least five years of formal education (mean = 11.47, SD = 4.77). RESULTS: A stepwise multiple linear regression model showed that, for most tasks, the number of years of education, age and FRWH were better predictors of performance when analyzed together rather than separately. In separate analysis, education was the best predictor of performance in language tasks, especially those involving reading and writing abilities. CONCLUSION: The results suggested that the number of years of education, age and FRWH seem to influence performance in the MTL-BR, especially education. These data are important for making diagnoses of greater precision among patients suffering from brain injuries, with the aim of avoiding false positives.


INTRODUCTION
Interpretations on the findings from neuropsychological assessments of language tend to have significant bias because of difficulty in distinguishing between cognitive effects due to brain damage and biological and sociocultural traits, in patients examined. 1,2 Thus, there is growing concern about understanding how age, gender, race, education and socioeconomic status, among other factors, may interfere with cognitive functioning. [3][4][5][6] Among the abovementioned factors, education level and age are the ones highlighted in the literature as the core influences on cognition. Overall, old age and low education level have been correlated with decreased performance. [7][8][9][10][11] However, this relationship is not always linear, given that there may be interactions between these factors, such as in naming and verbal fluency tasks, in which the effect of age is canceled when individuals are highly educated. [12][13][14] In addition, although education is considered to be the cultural factor that has the greatest influence on cognition, there are limitations in analyses on this variable. Education level is generally based on the number of years of study, which addresses differences in the duration and not the quality of education.
One alternative that has been suggested is to use reading level 15 as a predictor of cognitive function. Since participants who have higher frequency of reading and writing habits (FRWH) are more proficient in these skills and therefore score better in cognitive and language tests, 6,10,16-18 reading performance has been regarded as equally or more important than education level. It is also worth noting that the practice of writing can influence reading comprehension: people who read more write better and vice versa. 19,20 The effects of age and education level on healthy samples have been investigated in Brazilian research studies using the two main language assessment batteries: Boston Diagnostic Aphasia Examination (BDAE) 21 and Montreal-Toulouse Language Assessment Battery. 22 These studies have aimed to demonstrate the strong impact of these two factors on tasks involving oral and written comprehension, reading aloud, spelling, naming, written naming, reading numbers and copying. The results have revealed that older individuals with lower education levels are the ones who underperform. However, no studies investigating the independent effects of age, education and FRWH and their interactions in aphasia batteries have been conducted in Brazilian populations. 8,9,23 Several tools designed to assess language among patients with aphasia have been described in the international literature. The BDAE, 21 Aachen Aphasia Test (AAT) 24 and Western Aphasia Battery (WAB) 25 are the ones that stand out. Although not as widely used, another battery of Francophone origin that is used for language assessment is the Montreal-Toulouse Language (MTL) Assessment Battery. 22 The Brazilian Portuguese adaptation of this battery (MTL-BR) 26 has been seen to present interesting advantages with regard to evaluation and clinical interpretation of the different components of oral and written language. In particular, this adapted version has made it possible to analyze dissociations between different inputs and outputs, and different levels of complexity (word, sentence and discourse). 26 The psychometric measurements were verified in a previous study 27 that found evidence of validity (Cronbach's alpha 0.79-0.90) and reliability (mean test-retest score of 0.52) for the battery. Nevertheless, there is still a lack of research on healthy participants, with the aim of comprehending the effect of each sociocultural and/or biological factor on linguistic and language-related abilities.

OBJECTIVE
In this context, this study aimed to investigate the relationship between the Montreal-Toulouse language assessment battery and education, age and FRWH, using a sample of neurologically healthy adults.

Participants
The study included 233 neurologically healthy adults from southern Brazil: 151 females and 82 males. Their ages ranged from 19 to 75 (mean = 45.04; standard deviation, SD = 15.47) and education level ranged from 5 to 23 years of formal schooling (mean = 11.47; SD = 4.77).
The exclusion criteria included: a) impaired vision and/or hearing that was not corrected by means of visual and/or hearing aids; b) signs suggestive of neurological/psychiatric conditions; c) signs of moderate to severe depression (scores above 19 points), as measured using the Beck Depression Inventory (BDI-II); 28 and d) signs of cognitive decline, as measured using the clock-drawing test 29 in association with the Mini-Mental State Examination (MMSE), 30 with scores below 22 points for individuals with 5 to 8 years of schooling and below 24 points for more than 8 years. 31 In addition, participants who had a history of alcoholism and/or current or previous abuse of illicit drugs or benzodiazepines and antipsychotics, except for atypical neuroleptics (data collected through a questionnaire on the sociocultural aspects of health), 32 were not included in the study.

Materials and procedure
The participants were recruited from university environments and work centers (convenience sample). After receiving clarifications about the study, all participants signed a consent form and participated as unpaid volunteers. It is important to state that all ethical procedures were respected, with a guarantee that participation in the study would be voluntary. The study was conducted under approval by the research ethics committee of a higher-education institution (Pontifical Catholic University of Rio Grande do Sul; under approval no. 04908/09).
The FRWH was evaluated using an inventory that included questions about reading habits (magazines, newspapers, books and other materials) and writing habits (text messages, letters and other materials), and the frequency of each activity was scored as follows: 4 points for every day; 3 for several days a week; 2 for once a week; 1 for rarely; and 0 for never, with a maximum frequency score of 28 points. 33 In this sample, the 14-point band was regarded as the median. Scores higher and lower than 14 were thus denominated high FRWH or low FRWH, respectively.
Finally, the participants were evaluated using the MTL-BR battery. 26 This tool enables characterization of the subjects' oral and written language expression and comprehension behavior.
The tasks used for this study were: 5. Written comprehension: Assesses the ability to identify the input from visual images corresponding to words and written sentences. The task consists of a total of 13 items, five words (plates with six stimuli comprising one target and five distracters: one orthographic, one semantic, one visual and two neutral) and eight phrases (both simple and complex). The maximum score is five points for words and eight points for phrases, with one point for each correct answer.
6. Sentence copying: Assesses the ability to recognize and reproduce letters. The task consists of a sentence made of eight words. The maximum score is eight points, with one point for each word spelled correctly. Verbatim or servile copying is not considered to be correct.

Statistical analysis
The data were analyzed using SPSS 17.0. Initially, Pearson correlation analysis was used to investigate the relationship between age, education and FRWH and the scores from the MTL-BR tasks. Next, multiple linear regression analysis using the stepwise method was performed in order to identify which variables provided the best explanatory models. A significance level of P ≤ 0.05 was applied.

RESULTS
The scores from 17 out of the 22 tasks were significantly correlated with education, while the scores from nine subtests (oral narrative task, written comprehension, dictation, reading, semantic and phonological verbal fluency, written naming, written narrative and calculation) achieved moderate positive correlations. Ten tasks were negatively correlated with age (directed interview, auditory comprehension, oral narrative task total number of words, written comprehension, repetition, semantic verbal fluency, naming, written naming, written narrative total number of words and oral text comprehension). Performance in the oral narrative task, dictation, reading, semantic and phonological verbal fluency, written naming, written narrative and calculation presented significant moderate positive correlation with FRWH. The highest correlation was found between the written narrative task (total number of written words) and education variables, followed by the association between semantic verbal fluency and education ( Table 1). show, education was a significant predictor (P < 0.01) for most tasks, with explanatory power ranging from 6% to 28%.
Interestingly, it was also the only significant predictive factor in seven subtests: automatisms (form), oral narrative task (total number of IU), reading, number dictation, reading of numbers, written narrative (total number of scenes) and written text comprehension. The FRWH was a significant predictor only for automatisms (content), while age was the only predictor for the guided interview.
In general, among the tests in which education was combined with other variables, the variable FRWH contributed 2-3% to the explanatory model. In contrast, age contributed 3-6% to the explanatory power when combined with education in some tasks.
With regard to the explanatory models for each dependent variable, education level and FRWH jointly explained 32% of the variance of performance in the semantic verbal fluency and written narrative task (total number of words) and 30% of the variance relating to dictation. Education and age explained 25% of the scores for written naming. These three variables combined showed the greatest explanatory power (33%) for the semantic verbal fluency task. These findings corroborate the results presented in Table 1 (significant correlations).

DISCUSSION
The results from this study revealed that the MTL-BR is mainly influenced by education, such that higher education level was correlated with better performance. Likewise, other studies have shown the importance of the number of years of formal education in language tasks, with findings similar to those observed in the present study. 1,6,[8][9][10]33 As seen in the results, in this study there was a moderate positive correlation with education in tasks that involve dictation, written naming, written comprehension, auditory comprehension and written narrative. These tasks require reading-writing abilities that are developed over the course of education. However, education was not a predictor of the tasks relating to the guided interview, repetition, automatic speech (content) and oral narrative task (total number of words) ( Table 2). Previous research has shown that tasks involving graphic stimuli (written comprehension, dictation, reading, written naming, number dictation, reading of numbers, written narrative and written text comprehension) tend to be more sensitive to the influence of education. 6,8,11 The importance, albeit weaker, of FRWH in linguistic performance deserves to be highlighted, given that it increases the predictive power of the education variable. Regularly practicing reading and writing can compensate for low education levels in linguistic performance, 17,33 and the quality of what is read is one of the biggest predictors of cognitive performance. 16  It has been shown that not only education level, but also reading scores are predictors in the verbal fluency test, naming task and comprehension. 10,17,18 This is mainly because anyone who reads and writes more often possesses a richer vocabulary, and makes use of attention, mnemonic and executive skills.
People who tend to read less may present cognitive decline and worse outcomes in reading tasks, especially the elderly.
The habits of reading and writing are also considered to be one of the factors that contribute towards formation of a cognitive reserve, thereby preventing the effects of aging on cognition. 36 The findings from this study are in agreement with results relating to the effects of age, in a study on performance in the subtests of oral narrative, repetition, verbal fluency, auditory comprehension (sentences), written narrative, reading, written text and sentence comprehension, in which older individuals had lower scores than younger subjects did. 37 Like in other studies with similar tasks and batteries, 1,6,10 age had little effect on task performance. Despite the decline in task performance found among older age groups, it has been shown that in narrative tasks the elderly have better scores and higher production of words and sentences to express pictorial information, 38,39 thus corroborating the findings in our study.
This is because the elderly tend to have more repetitions, although these are not necessarily off-topic. 40 In speech, it is possible that repetitions are intentional, produced in order to emphasize certain information, 41 or even to make time for organization of thought. 42 Moreover, they may be due to the changes in communication style that occur naturally as a result of aging.
During the guided interview, age was the only predictor, contributing 3% to performance. In conversations with autobiographical topics, elderly people may have more personal observations, with frequent remembrance of the past. Moreover, with the aging process, narratives tend to have a complex plot, with more episodes and better management of resources, so as to maintain the interest of communicative interaction by making linguistic and extra-linguistic adjustments. 43 It is important to pay special attention to the heterogeneity of communicative performance, which is greater in the elderly population. 44,45 Some limitations of this study should be taken into consideration in order to make better use of its findings. The fact that illiterate or functionally illiterate patients were not included may have contributed towards the predictive outcome of models, with a maximum of 28% for the education level variable. Since the effect of education is not linear in cognitive tests (mainly because of lower education levels), a difference of one or two years of schooling interferes with performance. 6 In addition, it should be pointed out that both education level and FRWH were used as quantitative measurements, thus not covering specific measurements of education quality or the quality of the material read and/or written. 18 Thus, additional studies are needed in order to analyze these factors, so as to better understand the influence of these variables on language performance.
These data are important with regard to making correct diagnoses among patients suffering from neurological injuries, because of the potential for avoiding false positives. In addition, the variables studied are important because they establish normative data for clinical populations, especially aphasic individuals.

CONCLUSION
These results show the influence of sociodemographic variables, especially education and its association with FRWH, among the MTL-BR tasks. Examiners should take these variables into consideration when evaluating language performance, especially in clinical populations.