These last two observations by Ferrari deserve additional attention since black people tend to concentrate more heavily in the northeast and in rural areas. Indeed, Barcelos (1992), through an analysis of educational opportunities in Brazil, viewed the superior situation of whites in comparison with blacks through the perspective of spatial distribution.

One summary characterization of the spatial distribution of racial groups in Brazil points to the following trends: brown people are in the urban areas of São Paulo, the whites predominate in the south and southeast. This fact is not irrelevant in terms of educational achievement of people of color. We know that urban areas and the south and southeastern regions have the best educational attainment; thus, spatial distribution certainly has some sort of effect on the achievement of different racial groups. (39)

• Barcelos (1992) asserts, with other researchers (Lamounier 1976; Rosemberg et al. 1985; Rosemberg 1992) that the rates of illiteracy among blacks are consistently higher than among whites with similar socio-economic profiles.

There is evidence, however, that the mere concentration of blacks in the northeast and in rural areas does not completely explain these differences in illiteracy rates. Within a particular physiographic region, even in the wealthy areas, a very evident racial difference persists. This difference was studied by Rosemberg (1992, 305) using data from 1989. She found that "among the 20 million illiterate Brazilians (10 years and older), 41 percent are black residents in the northeast, whereas the total percentage of black residents in the northeast represents only 20 percent of the Brazilian population." The author concludes, "In other words, as long as racial discrimination persists, illiteracy will not disappear..." (Rosemberg 1992, 305).

• e)Brazilian researchers conducting research in the 1950 (Lamounier 1976; Begler n.d., cited in Lamounier 1976, 20) observed that "the social group that offers the greatest similarities with blacks are not the Jews but are women, as Gunnar Myrdal noted a quarter of a century ago...." However, a tendency observed in recent studies (Rosemberg et al. 1982; 1985; Rosemberg 1992, 1993) points out a trend towards equalization of male and female literacy rates. A lower literacy rate observed in older women can be considered a holdover from the past. The tendency that is currently more common is an equalization of male and female literacy rates in the younger age groups; in fact, in some age groups women have higher literacy rates than man (Rosemberg et al., 1982, 1985; Rosemberg 1992, 1993). Thus, past analyses do not appear to represent current reality in the educational arena.

• Efforts to comprehend the evolution of illiteracy rates taking into account both gender and race have not yet got beyond timid attempts. A limited focus on current data (1980 census), the lack of supporting longitudinal data, and an analytical focus restricted to the verification of the cumulative effect of age, rural/urban location, and region of residence have characterized this research. Thus, using data from the PNAD 1987, Rosemberg (1992) asserted that black women of age 40 and older have the highest indices of illiteracy in the country.

• Various studies focusing on gender point out important associations between literacy and aspects of women's reproductive and productive life, including: participation in the labor market and earning a salarie (Paiva 1981); fertility (Szmreczanyi 1988); interracial marriages (Berquo 1991); infant mortality for the total population and by race (Silva 1982; Tamburo 1987); and participation in the political sphere (Costa 1991).

If such studies are extremely important because they attempt to link schooling with other aspects of social life, the analyses are oriented, as a rule, toward the perspective that education acts as a facilitator, a determined or variable intervenor in the social dynamic. However, a more focused analysis of the relations between illiteracy and race uncovers the possibility of altering the reasoning, focusing instead on the impact of the indicators of the reproductive sphere of life on the educational opportunities of whites and blacks. Such a focus permits an interpenetration of the categories of gender and race.

Based on these preliminary discussions, we can now identify points that we intend to (and are able to) analyze within these different historical periods about illiteracy in Brazil by gender and race. In the first part, we synthesize the current descriptive panorama of literacy in Brazil, based on the data published by PNAD in 1990. Then we describe and interpret the secular tendency of illiteracy in Brazil from the perspective of gender and race.

Illiteracy and Exclusion

In 1990, more than 21 million people in Brazil over the age of 7 declared themselves illiterate, corresponding to 19.6 percent of the population for this age group. This median index hides an extreme amount of variation: 0.4 percent of the people between 15 and 19 years old residing in the southern regions of the country and living with families whose median income is above two minimum salaries (a referential wage unit in the Brazilian context: currently one minimum salary equals $65); 84.3 percent of children between the ages of 7 and 9 residing in the northeastern region and living with families with less that one-fourth of a minimum salary (PNAD 1990).

The illiteracy that persists in Brazil is fundamentally a product of the exclusion of populations that are impoverished in terms of social goods, particularly education. Thus, illiteracy in Brazil has a strong connection to the variations in family income that persist in the different regions and age groups (Table 1). To live in the northeast or in the southeast does not affect one's opportunity to become literate if one's salary is above two minimum salaries; to be older than 39--among southerners and northeasterners--also does not affect the opportunity to learn to read and write, if the family incomes are high. However, to be poor in these regions affects one's chances of being literate. Thus, for all the age groups, the indices of illiteracy are worse in the regions where a greater number of poor people are concentrated.

In this sense, it is neither the geographical region nor urban/rural distinctions that determine in and of themselves the different indices of literacy, but rather the greater or lesser concentration of poor people in these areas; it is also worthwhile to note the higher or lower concentration of income in these geographic spaces. It is because of this that the illiteracy rates vary between racial groups and do not vary more between the sexes, excepting a small proportion of the population that is made up of older females. If it is true that the median personal income of women, in all racial groups and in all the national regions, is inferior to that of men from the same categories (Bruschini 1994), this difference tends to disappear when one looks at family income, which is not taken into account when comparing whites and blacks.

It is possible, then, to suggest that the differences in literacy between racial groups and between genders comes from different processes: the small difference between male and female literacy rates persists in older age groups, and can therefore be considered a legacy of the past; the intense difference that is observed between illiteracy rates of whites and blacks, for both genders, constitutes a legacy from slave history, augmented by current policies that reinforce social exclusion.

When comparing the two genders within each racial group, the differences in literacy rates are minimal. However, comparing the literacy rates between black and white men or between black and white women reveals significant differences (Table 2).

Between men and women, for the general population, and between each racial group, important differences persist in the older age groups. It is upon analyzing this relic from the past that one learns with clarity the secular tendency of literacy in Brazil by gender.

Gender and Illiteracy in the Censuses of 1872 and 1980

A comparison of the data collected since the 1872 census through 1980 about male and female literacy provides evidence of an evolution that has been almost perfectly parallel until the 1940s when there was a notable convergence (Table 3 and Figure 1). Differences in literacy rates between the sexes were maintained at a relatively high rate until 1940 (around 8 percentage points) and began to decrease after that date, even when considering that after 1940 women comprised a greater proportion of the population than men (possibly because of better birth and post-partum conditions).

The data suggest that after 1940, a relatively high number of women integrated themselves into the school system, thereby diminishing the differences in levels of schooling between the genders. The process of schooling for women, which increased after 1940, contributed to the reduction in the percentage of illiteracy in the country in the 1940s, a trend noticed by other researchers but attributed solely to the internal mechanisms of the school system (Ferrari 1985).

If the desegregation of data by gender already highlights the importance of focusing on gender and its diverse effects on social dynamics, analysis of this same data from the perspective of race further complicates the picture.

The analysis of the evolution of literacy/illiteracy by race demonstrates another situation that steers us in the direction of demographic analyses about reproduction, introducing again the discussion of gender. We go then, in steps, to describing what we observed initially.

The data contained in Table 4, which is illustrated in Figures 2 and 3, provides evidence that illiteracy did not afflict, during the last 40 years, different racial groups the same way. On the one side are whites and mulattos which had similar profiles: a slow decrease in indices of illiteracy and a gradual increase in the number of illiterates. As we affirmed, this tendency demonstrates that the nation is not capable of making literate each new generation. That is, the pace of educational development is slower than the growth of the population.

When comparing whites and mulattos (Table 4), one observes that the mulattos are the racial group that shows the greatest growth in the absolute number of illiterates. Thus, while the general growth rate of illiteracy was 39 percent between 1940 and 1980, this rate was actually 15 percent for whites and 70 percent for mulattos when desegregated (Table 5).

The same trend is not visible, however, among blacks who demonstrated a simultaneous decrease in the percentage and number of illiterates during this same period. We suggest that the decrease in the absolute number of black illiterates does not detract from the fact that one part of this population would have become literate but is "disappearing" in the general population for reproductive reasons. This argument will be analyzed in the following sections.

In order to explain the decrease in the absolute number of black illiterates, one hypothesis is that black illiterates between 1940 and 1980 might have "migrated" to another racial group. As Piza and Rosemberg (1994) assert, the classification of race does not enjoy great stability or constancy in Brazilian life; that is, it is possible to alter one's racial identity during the course of a lifetime. As Wood's (1991) study evidenced, a "migration" has been observed in the way that people declare their racial identity from one census to the other, principally between blacks and mulattos. Given that Brazilian society views whiteness as the norm, it is not surprising that the self-classification process, and a related intricate process of intersubjective and social construction of identity, would result in shifting racial classifications (Piza and Rosemberg 1994). As Charles H. Wood has argued so well,

the estimated changes in self-identification through time have important implications for a longitudinal study [our case here] of racial inequalities in Brazil. The discoveries made through investigations that focus on socio-economic situations of black people [can bring] these subjects to possible perspectives in light of the significant proportion of black individuals that can reclassify themselves as mulattos. (1991, 104)

These commentaries could be valid for the case in question--and we would then end this interpretive exercise here--if it were not for the fact that the least educated people, not the most educated, had "disappeared" from the black classification. The logic of a racial "migration" from one color category to another corresponds with the ideology of whitening in the sense that "money (and status) whitens." As we have seen, illiteracy in Brazil is a powerful indicator of social and economic exclusion, and does not constitute a "passport" to whiteness. Therefore, it does not explain the decrease in the absolute numbers of black illiterates.

Another hypothesis that can be explored refers to the decrease in the black population in general in the country, a phenomenon identified by demographers (Wood 1991; Berquo, et al. 1986). As Elza Berquo affirmed, "throughout the history of the Brazilian population in the last 40 years, we have observed a change in the configuration of what is referred to as the color variable." The mulattos have had the highest rates of median growth annually, followed by whites and blacks. The black segment was the only one which, between 1940 and 1950, had a decrease in population.

When we compare the growth in population to the growth in illiterates in each subgroup of races, we see that among blacks in the 1940-1980 period there was a small increase in general population and a decrease in the absolute number of illiterates (Table 5, Figure 3). That is, a reduction in the black population was not general but took place principally among illiterates.

Moving beyond the "migration" hypothesis to explain a decrease in the number of illiterate blacks, it seems justifiable to look for explanations in statistical evidence from the last 40 years that substantiates the claim that the number of illiterate blacks would be diminishing. The hypotheses for explaining this population decrease come from a combination of factors including differences by racial groups in life expectancy, infant mortality rates, and marriage and fertility rates (the most important reproductive factors). These factors can also be viewed through the lens of gender.

Various social indicators that have been analyzed from the perspective of race provide evidence that the social destiny of blacks and mulattos is very similar and that it is not often that mulattos actually occupy a better position than blacks with regard to social goods (both symbolic and concrete). Nelson do Valle e Silva (1992), using data from PNAD 1988, affirmed that "poverty in Brazil is not color-blind." Poverty has greater impacts on mulattos (36 percent of this population is below the poverty line) and blacks (30 percent) than on whites. This incidence of poverty among blacks can explain the differences in life expectancy observed between racial groups. As Charles H. Wood (1991) states, life expectancy rates of whites greatly exceed those of blacks and mulattos: "Between 1950 and 1955, life expectancy for whites exceeded that of blacks and mulattos by 8.8 and 9.0 years.... Between 1975 and 1980, life expectancy for whites surpassed that of blacks and mulattos by 6.1 and 6.9 years respectively" (99). Thus, with a Brazilian illiteracy rate that is highest among the poor and elderly, we can suggest that among blacks and mulattos there might have been a decrease in illiteracy rates. However, if poverty and median life expectancy rates allow for a distinction between whites and blacks, they are not factors that establish a difference between blacks and mulattos, because both groups experience similar negative effects. In this way, these explanations do not sufficiently explain the significant decrease in the absolute number of illiterates among the black population.

In the article, "Blacks and Whites Marrying in Brazil," Elza Berquo (1991, 115) offers some clues for understanding the reproductive dynamics of black women. A summary of her observations regarding nuptial patterns by racial groups finds that: black women marry later in life than do their mulatto and white counterparts; celibacy among black women is greater than among mulatto and white women; same race marriages are the norm but when interracial marriages do occur, it is more common that the man has darker features than the woman; and that widowhood among black women is more common than among mulatto and white women. More than demographic phenomena, these aspects of Brazilian marriage uncover a connection between gender and race that produces fewer opportunities for procreation among black women than among their mulatto and white counterparts. In light of the dominant ideology in the country that equates "whiteness" with an improvement in social status and position for women, men tend to prefer women that are whiter and "of better racial background." Thus, black women have even fewer chances to procreate and they give birth to children that tend to be illiterate. Indeed, fertility among black women in the period that we are studying revealed lower rates than among white and mulatto women (Table 6).

The reproductive characteristics of black women diverge even more on another important point: the rate of infant mortality whose mothers classify themselves as black is higher than cases where mothers identify themselves as white or mulatto (Tamburo, 1987; Table 7).

Infant mortality for black mothers is significantly greater than that for white and mulatto mothers, especially among mothers who say they have no education (Tamburo 1987; Table 7). The high rate of infant mortality among children of illiterate black mothers, associated with black women's relatively low possibilities for procreation (i.e., their low ranking in the "marriage market") and lower rates of fertility, could explain the decrease in the number of illiterate people among blacks. Children from this group that might increase the number of illiterates at later ages (e.g., five years and older) are eliminated from the census counts because they die at such a young age.

Thus, as a result of the marriage market, black women have less opportunity to marry than mulattos and whites, diminishing their chances of having children. When they do marry, they seem to have fewer possibilities of having children; when they have do have children, if they are illiterate women, there is a lower chance that their children will actually live to expand the ranks of illiterates.

When we find that there has been a decrease in the last few years in the number of black people who are illiterate which cannot be accounted for by an increased rate of literacy, there are important implications for educational policy. Indeed, macro-structural studies (Barcelos 1992) provide evidence that education without regard to access or the drop-out rate of blacks and mulattos in schools construct strong and similar barriers to these two segments of the population. However, in light of the suggestion that the number of illiterate black people would have diminished because children from those groups that have the least schooling tend to be most rejected from the system, it is possible to conclude that racial discrimination in Brazil has its most dire consequences on the black population.

The link between race and gender becomes increasingly important (and opens new questions) when one knows that the greatest educational barriers for poor black children occur during the initial years of schooling, where the teaching staff occupies the lowest professional status and consists primarily of white women following instructions, in many cases, from white men.

*Fulvia Rosemberg holds a doctorate from the University of Paris in psychology. She is professor of social psychology at the Catholic University of São Paul and is a senior researcher affiliated with the Carlos Chagas Foundation. Dr. Rosemberg works as editor of the education journal, Revista Cadernos de Pesquisa, and has published widely on child development, day care institutions, and gender and race relations. Edith Piza is completing a doctorate in social psychology at the Catholic University of São Paulo, where she is working on a dissertation exploring gender, race, and intergenerational relations in Brazil.

NOTES

1. This work was completed within a study group on race, gender, and age at the Pontifical Catholic University of São Paulo. Support for this research on women was provided by the Carlos Chagas Foundation/Ford Foundation. This article was translated from Portuguese by Pia Wong, a contributor to this volume.

2. The IBGE is the agency responsible for the realization, processing and distribution of the census which is taken every ten years, and for the annual National Research on Households (PNAD).

3. During the twentieth century, census information on race was collected of 1940, 1950, 1960, 1980, and 1991. The census of 1960 experienced various technical problems and is no longer being used. The 1991 census had not been analyzed or published at the time this article was written (November 1994).

4. In 1980, the author confirms that blacks and mulattos registered similar rates of fertility when compared with people without education.

Barcelos, Luiz Claudio. 1992. "Raça e realizaçao educacional no Brasil." Dissertaças, Sociologia, IUPRJ.

Bercovich, Alicia. s.f. Considerações sobre a fecundidade da população negra no Brasil. (mimeo)

_____. 1987. Fecundidade da mulher negra: Constatações e questões. Textos NEPO 11: 47-100. Campinas: NEPO/UNICAMP.

Berquo, Elza. 1991. "Como se casam negros e brancos no Brasil." Desigualdade racial no Brasil contemporâneo. Org. Peggy Lovell. Belo Horizonte: CEDEPLAR, 115-20.

_____. 1988. "Demografia da desigualdade: Algumas considerações sobre os negros no Brasil." Novos Estudos CEBRAP 21: 74-84. São Paulo.

_____. 1987. "Nupcialidade da população negra no Brasil." Textos NEPO 11: 8-43. Campinas: NEPO/UNICAMP.

Berquo, Elza, Alicia Bercovich, and Estela Maria Garcia. 1986. "Estudo da dinâmica demográfica da populaçao negra no Brasil." Textos NEPO 9. Campinas: NEPO/UNICAP.

Bruschini, Cristina. 1994. O trabalho da mulher brasileira nas décadas recentes. II Seminário Nacional: Políticas Econômicas, Pobreza e Trabalho. Serie Seminarios N^o 7/94. Brasília: IPEA. 3-26.

Carvalho, José Carmello B. 1984. "Aspectos metodológicos dos quesitos sobre instrução no censo demográfico." Censos consensos contra-sensos. Ouro Preto: Associação Brasileira de Estudos Populacionais. 149-162.

Costa, Albertina de O. 1991. "O acesso das mulheres a cidadania: Questões em aberto." Cadernos de Pesquisa 52: 35-39.

Ferrari, Alceu R. 1985. "Analfabetismo no Brasil: Tendência secular e avancos recentes." Cadernos de Pesquisa 52: 35-49.

Galvez, Maria Eugenia Letelier. s.f. Analfabetismo feminino en Chile: Resultados preliminares. Santiago de Chile.

Gatti, Bernardette et al. 1990. "Alfabetização e educação básica no Brasil." Cadernos de Pesquisa 75.

Hasenbalg, Carlos A. 1992. "Notas sobre relações de raça no Brasil e na América Latina". Nosotras latinoamericanas. Estudos sobre gênero e raça. Org. Heloisa Buarque de Holanda. São Paulo: Fundação Memorial da América Latina. 52-58.

Brazilian Institute of Geography and Statistics (IBGE). 1992. Anuário estatístico do Brasil 1992. Rio de Janeiro: IBGE.

_____. 1928. Censo demográfico 1920. Rio de Janeiro: IBGE IV.1-2.

_____. 1950. Censo demográfico 1940. Rio de Janeiro: IBGE II.

_____. 1956. Censo demográfico 1950. Rio de Janeiro: IBGE I.

_____. s.f. Censo demográfico 1960. Rio de Janeiro: IBGE I.

_____. 1973. Censo demográfico 1970. Rio de Janeiro: IBGE I.

_____. Censo demográfico 1980. Rio de Janeiro: IBGE I.4.

_____. 1990. Estatísticas históricas do Brasil: Séries econômicas, demográficas e sociais de 1550 a 1988. Rio de Janeiro: IBGE. Séries e statísticas retros- pectiva 3.

_____. 1991 PNAD 1990. Rio de Janeiro: IBGE.

Lamounier, Bolivar. 1976. Educação. Cadernos do CEBRAP 15: 14-22. São Paulo.

Malmquist, Eve. 1992. "Women and Literacy Development in the Third World: A Rapsodic Overview." Women and Literacy Development in the Third World. Ed. Eve Malmquist. Linkoping: Departament of Education and Psychology-Linkoping University; UNESCO; SIDA, 7-41.

Marcilio, Maria Luiza. 1974. "Crescimento histórico da população brasileira até 1872." Cadernos do CEBRAP 16: 1-26.

_____. 1974. "Evolução da população brasileira através dos censos até 1872." Anais de História. Assis IV: 115-137.

Paiva, Paulo de L.A. 1980. AA mulher no mercado de trabalho urbano." Anais de História. Encontro Nacional de Estudos Populacionais, 2. Aguas de São Pedro. São Paulo: ANEP, 1981, 852-95.

Piza, Edith and Fulvia Rosemberg. 1994. A cor nos censos brasileiros. São Paulo: PUC-SP e Fundação Carlos Chagas. (mimeo)

Rago, Margareth. 1985. Do cabaré ao lar: A utopia da cidade disciplinar. Rio de Janeiro: Paz e Terra.

Reis, Maria Candida D. 1991. "Tessitura de destinos: Mulher e educação." Diss. PUC-SP. São Paulo.

República dos Estados Unidos do Brasil. Ministério da Industria, Viação e Obras Públicas, Directoria Geral de Estatística. 1898. Decenseamento de 31 de dezembro de 1890. Rio de Janeiro: Oficina de Estatística.

Rosemberg, Fulvia et al. 1985. Diagnóstico sobre a educação da mulher. São Paulo: Fundação Carlos Chagas.

Rosemberg, Fulvia. 1992. "Gender Subordination and Literacy in Brazil." Women and Literacy Development in the Third World. Ed. Eve Malmquist. Linkoping: Department of Education and Psychology; Linkoping University; UNESCO; SIDA. 302-35.

_____. 1993. "Subordinação de gênero e alfabetização no Brasil." Alfabetização: Passado, presente e futuro. São Paulo: FDE.

Rosemberg, Fulvia, Regina P. Pinto, and Esmeraldo V. Negrao. 1982. A educação da mulher no Brasil. São Paulo: Global.

_____. 1986. Situação educacional de negros (pretos e pardos). São Paulo: Fundação Carlos Chagas.

Rosemberg, Fulvia, Edith P. Piza, and Thereza Montenegro. 1990. Mulher e educaçao formal no Brasil: Estado da arte e bibliografía. Brasília: REDUC/INEP.

Schwartzman, Simon, Helena Maria B. Bomeny, and Vanda Maria R. Costa. 1984. Tempos de capanema. São Paulo: EDUSP, Rio de Janeiro: Paz e Terra.

Silva, Lea M. da. 1982. "Greater Opportunities for Women Related to Population Growth." Belo Horizonte. Paper presented at the Conference on Women in International Development, Michigan State University.

Skidmore, Thomas. 1991. "Fato e mito: Descobrindo um problema racial no Brasil." Cadernos de Pesquisa 79: 5-16.

Szmrecsanyi, Maria Irene. 1988. Educação e fecundidade: Ideologia, teoría e método na sociología da reprodução humana. São Paulo: NUCITEC/EDUSP.

Tamburo, Estela Maria Garcia. 1987. "Mortalidade infantil na população negra brasileira." Textos NEPO 11. Campinas: NEPO/UNICAMP. 104-27.

Wood, Charles. 1991. "Categorías censitarias e classificações subjetivas de raça no Brasil". Org. Peggy Lovell. Desiqualdade racial no Brasil contemporâneo. Belo Horizonte: CEDEPLAR. 93-111.