Doing and undoing gender in rice business and marketplaces in Tanzania

Abstract This paper contributes to the gender-and-marketplace literature by exploring whether and how the ongoing, under-researched food-to-cash crop transformation of rice in Tanzania reinforces or challenges the “doing of gender”. We apply Acker’s “doing gender” framework, where gender is done by following normative conceptions and undone by challenging them. We analyze women and men’s everyday practices and relations in terms of identities, divisions, symbols and interactions. The empirical material includes observations, in-depth interviews and focus group discussions with women and men traders, farmers and key informants at two rice markets in Kyela, south-western Tanzania. We find that this transformation of rice has resulted in more processes of doing than undoing gender. Too, more women than men undo gender. Since men and masculinity are constructed as superior to women and femininity, this makes it more difficult for men to undo gender. The structures of the marketplaces also seem to influence these processes. Surprisingly, the old marketplace offers more avenues to undo gender, whereas the new, government-initiated marketplace reinforces the doing of gender. We conclude that this commercialization trajectory, including associated interventions, exacerbates rather than reduces gender inequalities. Future agricultural interventions should therefore consider both technical and social aspects to yield desired outcomes.

ABOUT THE AUTHOR Mesia Ilomo (Master of International Trade) is a PhD candidate at University of Dar es Salaam Business School, Tanzania. His research interests are in agribusiness, cross-border trade, and policy analysis. Lettice Kinunda Rutashobya (PhD) is a Professor at University of Dar es Salaam Business School, Tanzania. She researches in agribusiness and agricultural value chains, entrepreneurship and internationalization of firms. Esther K. Ishengoma (PhD) is an Associate Professor at University of Dar es Salaam Business School, Tanzania. She researches on microfinance, linkages between formal and informal institutions, and business development. Katarina Pettersson (PhD) is an Associate Professor in Social and Economic geography at the Swedish University of Agricultural Sciences, Sweden. She is a feminist researcher focusing critical entrepreneurship studies and rural policy. Johanna Bergman Lodin (PhD) is a Researcher at the Swedish University of Agricultural Sciences, Sweden. She researches agricultural intensification and commercialization trajectories in sub-Saharan Africa from gender perspectives.

PUBLIC INTEREST STATEMENT
Gender equitable agricultural development has become a critical part of the development agenda of many sub-Saharan African countries, since gender inequalities impede social and economic development. Due to cultural norms, men tend to control the trading of cash crops, while women control the trading of food crops. Our study explores how gender is performed in the context of rice in Tanzania, a crop transforming from food to cash crop. We study the ways women and men perform gender in rice business and market places. The results show that transformation of rice, accompanied with new markets and advanced processing technology, reinforces gender imbalance. Women continue to trade at old marketplaces with old technology juggling trading and family care responsibilities. Interventions to enhance women's access to technology and highvalue markets, as well as reducing their care burden, are required to tackle gender inequality and attain Sustainable Development Goal 5 by 2030.

Introduction
Gender equitable agricultural development has become a critical part of the development agenda of many sub-Saharan African countries (Bullock et al., 2018). This is because gender inequality is a major problem in most African countries as it hinders both social and economic development (Spring & Rutashobya, 2009). In agriculture, men tend to control the trading of cash crops, whereas women control the trading of food crops such as rice (Doss, 2002;Njuki et al., 2011). There has been increased attention on examining factors leading to such inequalities in various economic and social spheres (Amine & Staub, 2009;Langevang et al., 2018;Lusiba et al., 2017;Rutashobya et al., 2009), including business, agriculture and their resulting outcomes. However, the question of how women and men perform as normatively expected of a particular sex category-that is, "doing gender", in terms of gendered identities, interaction, symbols, and division of labour, location in space and power, has largely been overlooked. Similarly, the way gender is "undone" and reduces gender differences, particularly in rural areas, is under-researched. Understanding how gender is done and undone is of utmost significance, as it is related to the potential progress towards attaining gender equality.
Drawing on the theoretical framework on doing gender (Acker, 1990;West & Zimmerman, 1987), we explore how gender is done, or undone (Deutsch, 2007;Kelan, 2018), in everyday social interactions in the rice sector in rural Tanzania. The rice sector presents an interesting context in which to study these issues. Traditionally a food crop controlled by women, rice in Tanzania has slowly been gaining value and is transforming into a cash crop (Achandi & Mujawamariya, 2016). Between 2001 and 2011, rice consumption per capita grew by 24 per cent because of rapid urbanisation, as well as a growing population and middle class, and changing consumer preferences (Lazaro et al., 2017;Wilson & Lewis, 2015) 1 Previous research suggests that when a crop produced mainly by women gains a higher monetary value, men tend to crowd out the women and take over production and trading (Doss, 2002;Njuki et al., 2011). Similarly, men have been seen to be more active in the trading of high-value cash crops than of low-value food crops (Aberman & Roopnaraine, 2020;Njuki et al., 2011). In relation to rice in Tanzania, women seem to control trading of rice in rural areas where returns remain low, whereas men control trading of rice in lucrative markets, including urban and export markets (Nkuba et al., 2016;. These trends may negatively affect household welfare, as women spend more of their income on household food, health and education, than men generally do (Mandel, 2004). The effect may be widespread in Tanzania, where rice is the second most consumed crop after maize (Achandi & Mujawamariya, 2016;Lazaro et al., 2017;United Republic of Tanzania, 2019). Although research has shown a gender divide in agriculture in terms of who controls which crop or technology (Aberman & Roopnaraine, 2020;Doss, 2002;Quisumbing & Pandolfelli, 2010), how rice markets are gendered and how men and women are affected in the transformation of rice from food to cash crop remains scantly documented. This paper, therefore, aims to examine and understand the ongoing transformation and developments in the rice sector in rural Tanzania from a gender perspective. Specifically, we explore whether and how the doing of gender in rice trading is maintained, or challenged, in terms of identities, gendered divisions, symbols and interactions.
Our research sites were two marketplaces, namely Kikusya and Kalumbulu in the Kyela district of south-western Tanzania, which is known to produce the rice most preferred by consumers in the country. Women have historically dominated rice production and trading in Kyela, whilst men have recently started to join the value chain, including engaging in the retail business which used to be constructed as feminine (Nkuba et al., 2016). Kikusya, a newer marketplace initiated by the government in 2015, incorporates new rice processing technology, while Kalumbulu is older and relies on more small-scale rice processing technology.
The establishment of the new marketplace, the introduction of new technology and other developments in the rice sector may boost value addition by taking the processing facilities and markets closer to farmers (Khoza et al., 2019). They may also attract new players and affect the practices and interactions of women and men in the rice business. The literature suggests that new markets (Quisumbing & Pandolfelli, 2010), and more so lucrative markets (Andersson et al., 2016;Mudege et al., 2015), tend to attract and become dominated by men. Actors in advanced and lucrative markets need to have high financial capacity and strong productive networks to facilitate business. Women's participation in such markets tend to be limited because of financial constraints (Kuada, 2009), and patriarchal culture that blocks their opportunity to benefit from business networks Van Den Brink & Benschop, 2014). However, some women are able to exploit emerging business opportunities in similar ways to men (Mandel, 2004;Spring, 2009). This is especially the case when agricultural interventions consider both the technical and social factors and successfully address both women's and men's needs (Amine & Staub, 2009;Kantor et al., 2015). Although the doing gender framework explores gender at interactional level, it does not explicitly address structural constraints (Acker, 1990;Deutsch, 2007). However, its application enhances the understanding of persistent structural gender inequalities (Deutsch, 2007;Pierotti et al., 2018), something missing in previous studies.
This paper contributes to the understanding of the gendered transformations of crops, from food to cash crops. While benchmarking a newer market against an older market with old technology, the paper also contributes to the literature on gender and marketplaces by highlighting the implications of government and private sector initiatives to establish new rice marketplaces with improved technology in the rice sector. Further, we enhance the analytical value of Acker's (1990) doing gender framework by merging the doing gender and undoing gender perspectives, and update the visual tool introduced by Kelan (2018). The adapted analytical framework permitted a comprehensive analysis of practices of both women and men in the rice business and marketplaces in Kyela, Tanzania, which is largely an informal sector. Such an analytical approach and the knowledge gained from this kind of context are limited in extant literature, and the paper thus contributes to this. Additionally, we contribute to the literature on the engendering of value chains in agriculture. Our findings further show both progress and resistance towards gender equality, and can thereby contribute to achieving SDG 5.
The remaining part of the paper is organized as follows. First, the review of literature and outline of the theoretical framework are presented. Second, the methodology and research context are described. Third, the paper presents and analyses the results. Finally, it discusses the findings, concludes, and highlights some implications and avenues for further studies.

Doing and undoing gender: a literature review
Feminist approaches in research generally recognize male dominance in social arrangements and want to change this domination (Calas & Smircich, 1996;Pierotti et al., 2018). Thereby they are critical of the status quo and pursue social change to achieve fair and equal societies (Calas et al., 2007). There are various ways of defining and differentiating between different feminist approaches depending e.g., on their ontological and epistemological perspectives, including views on sex/gender. Calas et al. (2007) recognise liberal-, radical, psychoanalytic, socialist-, poststructuralist/post-modern-, and post-colonial feminist approaches, and cluster these approaches into two groups. The first group, liberal-, radical, and psychoanalytic approaches, ontologically assumes that women are disadvantaged because of their condition as women, and focuses on women's difference from men. Gender refers to specifically "sexed" bodies according to these approaches. In contrast, the second group, including socialist-, post-structuralist/post-modern-, and post-colonial perspectives, characterises gender as the product of power relations, developed from historical processes, dominant discourses and institutions. Gender is thereby understood as distanced from an individual's personal experiences, and as a constructed process and practice, based on specific linguistics, historical, cultural and political conditions. These approaches are used for analysing the relations of power in which gender (and other identities) are seen to produce, reproduce, or resist social systems.
The doing gender framework is pursuing a socialist feminist approach (Calas et al., 2007), and it emphasises that gender is performed (or "done") rather than assigned by biology (West & Zimmerman, 1987). In this framework, gender is captured through the practices of women and men in everyday interactions. Gender performance is subjected to constant assessment as to whether it fits masculine or feminine criteria and tends to emphasise the differences (Pierotti et al., 2018;West & Zimmerman, 1987). Acker (1990Acker ( , 1992 proposes four interactive but analytically distinct ways that gender is done at the workplace. These processes include: • The social construction of divisions of labour, behaviour, location in space and power; • The social construction of symbols and images that explain, express, reinforce or sometimes oppose those divisions or maintain societal masculinities and femininities; • Interactions, at individual or group level, including those "patterns that enact dominance and submissions" and "create alliances and exclusions" (Acker, 1992, p. 253); and • Gendered components of individual identity-the internal mental work of individuals as they construct what is (in-)appropriate for them as members of a workplace or society.
Women's and men's practices tend to take different values and are ranked differently (Evans, 2014;Pierotti et al., 2018). Masculine labour, symbols, interactional aspects and identities are frequently constructed as superior (Acker, 1990;Martin, 2003;Pettersson & Cassel, 2014). West and Zimmerman (1987) and other scholars, including Mavin and Grandy (2012), argue that gender is always done, and that gender change entails redefining normative conceptions, i.e. redoing gender. These scholars are pessimistic that gender equality can be achieved. In contrast, we are optimistic that gender equality is achievable through processes of undoing gender; i.e. through everyday interactional practices that reduce gender difference (Deutsch, 2007;Kelan, 2018). In exploring undoing gender, we focus on social processes through which the gender binary is challenged. Our approach to undoing gender is different from Butler (2004), who focuses on how the gender binary is discursively challenged. Women and men may undo gender along the same processes proposed by Acker (1990Acker ( , 1992. While processes of undoing gender may lead to more egalitarian societies, we are aware that these processes are complex and may involve resistance (Pierotti et al., 2018).
Empirically, various scholars have explored doing and undoing gender practices in workplaces, especially in developed countries. The gendered component of individual identity entails efforts to make sense of how one sees oneself and others in relation to being the same (Pullen & Simpson, 2009). The identity often sanctions the corresponding appropriate behaviours as a member of the workplace, e.g., choice of appropriate work, how to interact and how to present oneself (Acker, 1990). Normatively, men are constructed as breadwinners and superior, whereas women are constructed as carers and subordinates (Amoo et al., 2019;Deutsch, 2007). Men do gender by maintaining their identity as breadwinners (McDonald, 2013;Nentwich et al., 2013). Women manage care responsibilities as mothers and/or wives (Anderson, 2008). Consequently, women's geographical mobility is constrained (Mandel, 2004). Some women, for example, in the diner industry in the US, use their motherly identity when they interact with the college student clientele, whereas other women go into business with their children (Anderson, 2008). In Western contexts in particular, such as Sweden, women entrepreneurs have access to childcare facilities (Pettersson & Cassel, 2014), which, while necessary, are largely absent in Tanzania and Kyela, in particular (Langevang et al., 2018). Men might also assume the carer responsibilities. Unfortunately, change in this direction seems limited and may be reversed, as has been reported to be the case amongst some men in the Democratic Republic of Congo (DRC) (Pierotti et al., 2018). Thus, the ability to balance work and family affairs seems critical for women business owners (Gherardi & Poggio, 2001;Langevang et al., 2018).
Researchers have also examined the gendered divisions in developing countries. For example, Mudege et al. (2015) explored the influence of gender norms on potato trading in Malawi, where they found women exploiting local markets and leaving the distant, more lucrative markets to men. Similar experiences were found in the trading of cassava leaves in Tanzania (Andersson et al., 2016). Gendered divisions of labour were examined in farm tourism entrepreneurship in Sweden (Pettersson & Cassel, 2014), street vending in Mozambique (Agadjanian, 2002), the diner business in the US (Anderson, 2008) and livestock auction markets in the US (Pilgeram, 2007). These studies mainly found women and men doing gender by handling tasks constructed as feminine and masculine, respectively. In Tanzania, Nkuba et al. (2016) found women were engaging in rice retailing while men were collecting rice and engaging in wholesale rice business. There are limited but emerging undoing gender practices, such as women accessing distant markets in Benin (Mandel, 2004) and Zambia (Evans, 2014) as well as men handling housework in Democratic Republic of Congo (Pierotti et al., 2018). Also, some men were found to undo gender by engaging in the initial preparation of meals in the US diner industry (Anderson, 2008) and taking in paying guests in the tourism business in Sweden (Pettersson & Cassel, 2014). Further, Anderson (2008) found the work-home division was blurred in the US by entrepreneurs entertaining family members and friends at the workplace.
Symbols and images take different forms, including language, dress, skills, knowledge, ideology and financial resources or commitments (Acker, 1990;Pilgeram, 2007;Sayce, 2012). Men are constructed as knowledgeable, more skilled, rough (Acker, 1990;Deutsch, 2007), physically stronger (Lusiba et al., 2017) and financially stronger (Agadjanian, 2002) than women. Extant research shows that women do gender by dressing smartly at the workplace (Pilgeram, 2007) and equating business in the diner with providing food at home (Anderson, 2008). Andersson et al. (2016) found women cassava traders in Tanzania were presented as physically weak. Male hairdressers in the UK and livestock traders in the US did gender by appearing rough (Hall et al., 2007;Pilgeram, 2007) and male nurses demonstrated physical strength (McDonald, 2013). Pilgeram (2007) found some women, in a livestock auction market, were wearing dirty clothes and presenting themselves like men, and thereby undoing gender.
With regard to interactions, women are normatively constructed as caring, emotional, passive and soft-spoken, whereas men are normatively constructed as less welcoming, individualistic, rational and non-expressive (Deutsch, 2007;Martin, 2003;McDonald, 2013). Men in the US, Tanzania and the Netherlands, countries that are culturally different, were found to prefer interacting with fellow men (Martin, 2003;Rutashobya et al., 2009;Van Den Brink & Benschop, 2014). Martin (2003) and Van Den Brink and Benschop (2014) found men unconsciously, and sometimes consciously, distancing themselves from women because of the risk of their interaction being confused with intimacy. Further, women were often found doing gender through caring and nurturing their customers (Agadjanian, 2002;Anderson, 2008). Extant literature presents some undoing gender practices, such as women engaging in profane conversations (McDonald, 2013) and men caring (Hall et al., 2007) and deliberately identifying with women (Van Den Brink & Benschop, 2014). Hall et al. (2007) found men in real estate benefiting from showing their emotions, whereas men in firefighting devalued emotions.
Of all the reviewed studies, only Sayce (2012) and Pettersson and Cassel (2014) comprehensively studied gender using Acker's (1990) doing gender framework. However, Sayce did not cover undoing gender. The adapted doing gender framework permits exploration of the relational aspect of the four ways of doing and undoing gender. Further, doing and undoing gender processes seem to vary by sector, while none of the studies explored the processes in agricultural crop businesses and marketplaces. Additionally, the doing and undoing gender studies in developing countries, known for constraining socio-cultural factors, are scant but necessary to generate knowledge and inform policies. Such studies are especially needed in women-dominated sectors, such as the rice business in Kyela, which are under-researched compared to men-dominated sectors (Pullen & Simpson, 2009). Empirical studies on women-dominated sectors tend to study men, whereas few cover both women and men.

Study site
The Kyela district is located in the Mbeya region, one of the six major rice producing regions of Tanzania. The district is well-known for producing the best and most aromatic rice in the country, where Dar es Salaam, the major market, accounts for 60 per cent of the total domestic rice consumption (Lazaro et al., 2017;. In 2012, Kyela's population comprised 106,012 men and 115,478 women (United Republic of Tanzania, 2017), with the majority working in farming and agribusiness. Rice, the district's most important food crop and second cash crop after cocoa (United Republic of Tanzania, 2017), is traded mainly in designated marketplaces. Kyela has five major rice marketplaces, namely Kalumbulu, Kikusya, Ipinda, Ngyekye and Kasumulu. The trading at Kikusya is controlled mainly by men, whereas women control the trading in the other marketplaces. While Kalumbulu and Kikusya are in Kyela town, Ipinda, Ngyekye and Kasumulu are located in rural Kyela. Maps of the study site, including the selected marketplaces, are presented in Figure 1. The basis for selecting case studies for this paper is further discussed in the sampling and data collection section.

Sampling and data collection
The paper is based on 19 in-depth interviews with 11 traders and farmer-cum traders, and eight key informants, as well as four focus group discussions (FGDs). Three FGDs included five to eight women and/or men traders, and one had eight key informants. Two FGDs with traders were gender-disaggregated with eight members each, while one was mixed, comprising two women and three men. Gender-disaggregated FGDs, conducted at a ward hall in Kyela, were effective in collecting gender-related data, especially in this patriarchal community, where men may otherwise suppress women's voices. The mixed FGDs with traders took place at one of the trader's offices at Kalumbulu and served as a means for accessing traders for further interviews. Further, the FGD with key informants included five women and three men and took place at the Kyela district council hall. In total, the FGDs included 15 women and 14 men traders and key informants. We also made approximately 15 visits over four months to observe the market operations and practices of market participants at the marketplaces. We used multiple sources and methods of data collection, as some gender practices are unintentional, or cannot be verbally expressed by participants, and/or women and men are unaware of them (Kelan, 2018;Martin, 2003).

Marketplaces
Two marketplaces, Kalumbulu and Kikusya, both in Kyela town, were sampled ( Figure 2). The markets differ in terms of background, participant composition and rice processing technology. Women account for the majority of rice traders at Kalumbulu, while the majority of rice traders at Kikusya are men.
Located in the centre of Kyela town, the Kalumbulu market was established in the mid-1980s with diesel-powered rice processing facilities. It currently includes 21 electric-powered rice processing facilities, some designated rice drying areas and some rice shops. The processing facilities are small, and mainly process rice without grading. In 2018, however, one small rice processing and grading machine was installed. The market has about 200 traders and 800 other players: processors, machine operators, supervisors and manual workers, loading or offloading trucks. Trading is conducted within the processing facilities, in open areas and business premises which sometimes involve a sharing arrangement, especially among women. The marketplace is surrounded by general shops, residential housing and street food vendors.
The Kikyusa market is located at Kikusya, 3 km from Kyela town, along the road from Kyela to Mbeya/Dar es Salaam and the Malawi border. It was established in 2015 by the Kyela district council, who installed a processing facility to boost the value addition on rice in the district. Later, a private investor installed a slightly more sophisticated processing facility along with a small ordinary rice processing machine to cater for the needs of customers with small consignments. The large, advanced processing technology installed at Kikusya enables the rice to be processed, sorted and graded. Therefore, unlike Kalumbulu, Kikusya market sells only graded rice. The grading means the average price of rice at Kikusya market is usually higher than at other markets in Kyela. The marketplace also has far fewer players (approximately 20), including traders.
In terms of infrastructure, Kikusya has two buildings: one for the private investor's processing facility, and the other a wire fenced facility housing the government processing machines. Business is done mainly within the processing facilities. There are few residential properties near the marketplace.
Farmers or sellers often deliver unprocessed rice at the marketplaces. Unprocessed rice supplied at Kalumbulu originates from different parts of the district, and occasionally from Malawi, whereas rice traded at Kikusya mainly originates from Kikusya village. The private investor at Kikusya, who is determined to promote Kyela rice, discourages rice imports. Kikusya occasionally grades rice processed in other areas. Processing rice provides three products: quality rice, broken rice and husk. At Kalumbulu, a trader or a farmer has the rights over all three products. Broken rice and/or husks are often used to pay people hired to dry and process the rice who, in turn, sell the broken rice for a lump sum or use it to start a retail rice business. At Kikusya, husk is retained by the processor and drying labour is paid in cash. The daily volume of trade during the peak business season (June to August) at Kalumbulu and Kikusya is up to 75 tons and 6 tons of processed rice respectively.

Study participants
Participants from both marketplaces were selected on the condition that they were engaged in the rice business and were willing to participate in the study. The ward and rice market officials facilitated the selection of participants for interviews and FGDs. The mix of participants reflected the composition of traders at the marketplaces and the respondents' knowledge of the rice business and/or gender. The selected participants included both market participants and key informants. The traders and farmers interviewed in-depth were in their mid-30s to mid-50s and each had between two and six children. These participants were engaged in the rice business in different ways. Table 1 presents their pseudonyms and profiles.

Fieldwork
Data collection was mainly done by the first author and a woman assistant. It started with a familiarisation visit in July 2017 and comprised observations, interviews and FGDs, which were conducted from January to April 2018 after securing research approval from the government of Tanzania. Interview and focus group guides facilitated the interviews and FGDs. They covered different topics, including gender in the rice value chain, women's and men's participation in rice trading, gender norms and practices, and the dominance of women in the rice business. Interviews lasted between 10 and 60 minutes, while the FGDs took between 55 and 90 minutes. We recorded all the in-depth interviews and three of the FGDs, cumulatively making 500 minutes of recordings. Traders in the mixed FGD rejected recording because they were unsure of the details of discussion. We conducted and documented or transcribed all interviews and FGDs in Swahili. The verbatim transcripts totalled 90 pages. The first author later translated the selected materials from Swahili to English. The quotes included in the paper were selected on the grounds that they are considered representative of the doing and undoing of gender practices in the rice business and marketplaces.

Data analysis
Data analysis began during fieldwork through daily reflections following the interviews, FGDs and observations. However, we conducted the thorough analysis after the fieldwork. We first reviewed the transcripts, notes from unrecorded interviews and FGDs, field notes and contextual information, and interview summaries three times. Then we used the NVivo 12 Pro qualitative data software to code and cluster transcribed selected materials useful for this paper into the four doing gender processes: identities, gendered divisions, symbols and interactions. Within each process, we identified women and men doing and undoing gender practices in the rice business and marketplaces in Kyela. The identification of these practices forms the basis for answering the question: how gendered interactions are maintained, or challenged, and how the market structures and systems influence these processes. Some of the practices are informed by previous literature, whereas others emerged from the data. The first author did the initial analysis and worked with the four co-authors to produce the final results reported in this paper. We did the analysis with the audiotaped materials running to ensure that participant accounts were accurately presented, which also allowed us to explore other aspects such as the pitch and tone in conversation.

Results
This section presents findings on the doing and undoing of gender practices at the two marketplaces in relation to gendered identities, gendered divisions, symbols and interactions (see Table 2 for an overview). 2 Note: M stands for man and W for woman. All respondents except Flora (who has secondary education) are primary education holders. All respondents are also farmers.

Identities
We found women and men making sense of what is (in-)appropriate for them in terms of work, attire, behaviour, language and means of interacting with a person of the same or opposite gender category. They were "creating the correct gendered persona and hiding unacceptable aspects of one's life" (Acker, 1992, p. 253). Women's and men's practices sometimes challenged the normative gendered identities.
Masculinities were frequently linked with work and less so with care responsibilities. Women demanded that men honour their paternal identity by guiding and mentoring children, especially in education. However, some men, for reasons including e.g., excessive alcohol consumption (cf. Amoo et al., 2019), fail to honor the paternal role, which is then taken on by their wives. In line with the construction of good mothers, women could not leave children unattended or allow them to miss out on their basic needs. Further, women and men in FGDs tended to mark non-working married women as "bad wives." Consequently, many women in Kyela engage in incomegenerating activities, and rice traders in the district are often women. The rice marketplace helps women to build confidence to the level of outcompeting men in business. The focus group participants noted that the rate of women joining the marketplace was higher than that of men. It was also clear from the mixed FGD with traders that men are exiting the Kalumbulu marketplace. Men may feel uncomfortable working at a women-dominated marketplace. However, study participants in Kyela argue that the proportion of men in the rice business today is higher than that of the mid-1980s. Men sometimes justify their participation in the sector based on physical strength, financial strength and/or freedom of mobility, similar to men in other womendominated sectors (Andersson et al., 2016;McDonald, 2013). Principally, both women and men made sense of who they are and who they are not as members of society and the rice marketplace.
Further reflection on the internal mental work of women and men in the rice business and marketplaces in Kyela is included in the detailed analysis of the other three gendering processes: gendered divisions, gendered symbols and gendered interactions.

Divisions along the gender lines of labour, power and locations in physical space
The analysis revealed gendered divisions. While both women and men participated in the rice business, their involvement was different.
We found women to account for the majority of rice business entrepreneurs, with most doing business at designated marketplaces, and often close to their homes. Anderson (2008) argues that proximity of the workplace is key to enabling women to handle care responsibilities. Women accounted for about three-quarters of rice businesses at Kalumbulu. Women rice traders there included retailers, brokers and agents for large buyers from different parts of the country and beyond the national borders. Retailers buy rice from farmers and other sellers to sell to end consumers. Brokers link the sellers and buyers of rice for commission. Rice agents buy rice on behalf of other traders, often doing rice business outside the district. While women have been rice agents for a long time, men are increasingly becoming agents. The few men at Kalumbulu mainly offered brokerage services. Discussing men and women's participation in business at the marketplace, the marketplace chairperson (a woman), stated: "We are in the same business, rice business . . . the difference is that women buy large volumes of rice. Men are mere brokers and do not buy rice. They can link me with customers." In contrast, men account for the majority of traders stationed at Kikusya. The price of rice there is usually higher than in other markets because the rice is graded. Thus, men control the market for the highly valued product that is graded rice. Similarly, Mudege et al. (2015) found men controlling the trading of seed potatoes in Malawi, whereas women traded the low value ware potatoes. Some men are reluctant to allow their wives to engage in rice business at Kikyusa, where the majority of participants are men. Yet, despite the objections of her husband (Gabriel), Jane participates in rice business with her brother-in-law: "It is my brother-in-law who often goes to the marketplace to sell rice." In contrast to women staying put at Kikyusa, many men, and a few women, rice traders travelled to villages to buy unprocessed rice. This practice sometimes assured them of quick financial benefits, through high profits. The following is deduced from David, a man trader: "As you know, when a man goes to the village, because he is powerful . . . buying unprocessed rice and taking it to town [Kalumbulu] gives him more profit than staying at the marketplace." In addition, for couples with both partners in the rice business, the husband was often engaged in buying rice from villages, whereas the wife sold the processed rice at the market. This is also consistent with experiences in other related women-dominated sectors such as street food vending, where men source the products and let their wives serve customers (Agadjanian, 2002).
The leaders at both marketplaces handled conflicts among market participants, coordinated players in case of problems and ensured moral behaviour. Normatively, men hold leadership positions, while women are subordinated. Yet we found that women mainly led Kalumbulu, the largest marketplace in Kyela, where they managed, among other things, to restore peace following a couple of incidences of young men using abusive language and destroying sacks of rice. In contrast, men led Kikusya, which is consistent with the normative conception. The entry and exit of buyers at Kikusya was controlled, as women traders in particular were largely discouraged at the marketplace, meaning there were more opportunities for men than for women. It has been noted that assuming a leadership position is a strategy employed by men to justify their presence in women-dominated sectors (McDonald, 2013); we find this at Kikyusa.
Men were expected to provide for their families as breadwinners, and were therefore frequently committed to work to earn an income. They entrusted the care responsibilities to women. Radson, a rice trader and supervisor at Kikusya, serves as an example. He arrived at the marketplace at 6 am and sometimes left for home after midnight. Whilst he put aside rice for household consumption and ensured that the family had enough food, he left the care work to his spouse. Radson explained: "If they [family members] misuse and use up all [rice], they cannot blame me but I am still a father, I have to take responsibilities because the children have nowhere to go. I have to rescue the situation." Many women were happy their husbands assumed the role of provider and expected the men to honour it. One was Hellen, a farmer and the wife of a rice trader at Kikusya, who described the family work schedule: "When we arrive home from the farm, I prepare his meal, he takes a shower and leaves for business." However, some men could not provide for their families and some households did not include men, or included men who were alcoholics. Some women, therefore, are becoming de facto breadwinners and seem to be as dedicated to work as men. Nevertheless, productive roles do not exempt women from care responsibilities. For example, women retailers doing business in the evening at Kalumbulu seemed to be bound up in housework, limiting their possibilities to work at the market (cf. Bergman Lodin et al., 2019). Asinati's daily schedule often started with cleaning the house and preparing breakfast before going to the nearby Kalumbulu market to buy rice. She then went home to prepare lunch before returning to the market for business. Other women also planned their daily schedule to accommodate both business and care responsibilities. In addition, some women traders occasionally took their children to the marketplace. The proximity of the business to the home and logistical efficiency are key to making the care responsibilities and engagements in rice businesses work (cf. Anderson, 2008). Working women juggling multiple roles seems to be a global phenomenon which hardly changes (Amine & Staub, 2009), although assistance from family labour may sometimes help (cf. Anderson, 2008). However, only a few men assumed the care responsibilities in Kyela.
Despite the reproductive roles ahead of them, some women were fully dedicated to the rice business to complement the income of their husbands. Rose, a rice trader married to a civil servant, who does business at Kalumbulu, said: "I leave the marketplace at 7 pm because he should be able to tolerate that. It is the two hundred I make [implying small income]. His salary comes at the end of the month. I take rice with me when I go home and bring vegetables on daily basis. I have to plan my time. He understands." In addition, some women traders had to arrive at the market as early as 5 am, at which time it is usually dark in Kyela, in order to meet the high competition. Also, due to problems with the electricity supply and/or pressing demands, both women and men at Kalumbulu sometimes worked up to midnight. The women were then escorted home by their husbands or sons to reduce the risk of physical attacks and sexual harassment. Women were aware of what was expected of them and that being escorted by men sometimes legitimised travel during the night, which is otherwise constructed as unacceptable for women in Tanzania (Andersson et al., 2016), as well as in other countries such as Nigeria and Kenya (Bergman Lodin et al., 2019). Despite living in the village, women at Kikusya did not arrive at the marketplace as early, or stay as late, as men, since they accounted for a minority of traders, but they made up the majority of rice drying labour. Radson noted that women were "aware that rice drying starts at 8 am" and thus arrived after the men. Also, rice business at this marketplace took place within processing facilities which were controlled by men. Thus, men often controlled the starting time and the end of the business day.
Most men were not happy to see women failing to handle care responsibilities because of their participation in business. One example is Radson, one of the men agents and retailers at Kikusya, who noted that his wife did small business which entailed some travelling. He accepted her not preparing meals for him, but "she must make sure that my children are fed." Radson considered his wife to be "misbehaving" by engaging in her business activities, implying that she no longer qualified as a good wife. In contrast, David, one of the occasional traders at Kalumbulu market, had gradually accepted his wife's reduced involvement in the housework. The need for women to handle care responsibilities was also emphasised by the Kalumbulu chairperson during the focus group discussion, albeit criticism from fellow women traders was aired. Thus, men and women, including leaders, resist transformation of family care responsibilities. This may be challenging for women, especially those trading at Kalumbulu, who have no guarantee they will have customers. It may also limit women's chances to access distant markets, which tend to be more lucrative (Mandel, 2004).
Other divisions of labour generally followed hegemonic normative conceptions, with women handling manual tasks and support services, including drying, bagging and polishing rice, in much the same way as in many parts of South Asia where technology is also underdeveloped (Harriss-White, 2005). Men performed technical tasks, such as operating rice milling machines and machines for zipping rice bags. Radson at Kikusya said that women "worry about the skills needed for using the sewing machine. The machines are heavier for them. It requires masculine energy to sew [the rice bags]." In addition, men performed tasks requiring upper body strength, such as loading and unloading trucks and moving unprocessed rice in carts from drying areas to processing facilities, especially at Kikusya, similarly to tomato trading in Tanzania (Khasa & Msuya, 2016). Andersson et al. (2016) also found the construction of women as physically weak, which denied women access to lucrative markets of cassava leaves in the coastal zone of Tanzania. Men in women-dominated sectors such as the rice business in Kyela thus perform "masculine" tasks as a strategy to negotiate their gendered position (cf. Nentwich et al., 2013).

Symbols and images
We also found practices related to gendered symbols in the form of a dress code while conducting rice business. Similar to practices in some contexts such as Uganda (Lusiba et al., 2017), "good women" are expected to dress smartly and cover their body parts. Elsewhere, women also tend to despise fellow women who dress too sexily at workplaces (Carr & Kelan, 2016;Mavin & Grandy, 2012). Society produces norms about dressing because of the tendency to link women with sexual propriety (Boyer et al., 2017). Accordingly, the majority of women traders in Kyela went to the marketplace with handbags holding their Kitenge (a loose cloth worn by women in casual work and social events, symbolising respectability), and once there they put it on. The Kitenge was also viewed as protecting them from the dust produced when measuring rice. Kitenge also served as a backup in case their dresses were damaged when they bent down to dry, measure or bag rice. Women removed the Kitenge, which was dirty, as they exited the marketplace. They still, however, had to appear respectable to reduce the risk of sexual harassment while on their way to and from the workplace, which sometimes happened during the hours of darkness. The need for appropriate attire among women, especially at night, has also been reported in countries such as Australia (Boyer et al., 2017), Nigeria andKenya (Bergman Lodin et al., 2019). It is interesting to note that men engaging in sexual harassment may also encounter health risks (Amoo et al., 2019).
Men's dressing practice was different, as they did not change their clothing on arrival or exit. That may be linked to several reasons. Men in Kyela are not confronted with the dust and seldom touch rice. Commenting on this, Rose, a woman trader at Kalumbulu, said: "Men are mere brokers, they rarely handle rice and face the dust. They also do not care how they appear during and after business. We [women] care about it." Notably, while men engaged in all kinds of activities around rice at Kikusya, none changed their clothing. This makes the following reason relevant: men were expected to report home in the same clothes-otherwise, they had to explain why and where they changed, indicating a suspicion they were engaging in illegitimate affairs. Our findings corroborate Hall et al. (2007), where men in hairdressing in the UK appeared "rough" to maintain their masculinity.
We also found gendering of the tools used in ascertaining the quantity of rice. Reference to volume and/or weight measures was common in interviews. Historically, both women and men traders used volume measures such as buckets (in wholesale) and bowls (in retail). At present, the district government discourages volume measures and encourages weight measures in the rice business through the Weights and Measures Act (Chapter 340). Traders, both women and men, running businesses of different sizes, are expected to switch to weight measures. However, the price quotation in Kyela at the two marketplaces is still often given in terms of 20-litre buckets (often equated with 20 kilograms). Many traders in the rice business also use volume measures. Volume measures were more common at the Kalumbulu marketplace, where the majority of traders are women. For example, we saw several women retail traders seated outside processing facilities and in open areas selling rice using volume measures. Among them are women retailers trading rice at the evening market. One of them, Asinati, said: "I started three years ago. I was selling vegetables, then I started buying a bucket of rice and broken rice. I use a bowl and do the rice business on tables." Traders such as Asinati may be financially unable to afford weighing scales. According to the informants at Kalumbulu, the change from volume to weight measures is less costly for men, as the majority of them are brokers.
The use of weight measures is more common at Kikusya marketplace. This marketplace mainly targets large and commercial buyers, and thus mainly uses the industrial weighing scales (costing about USD 735) available there. Yet men traders occasionally use bowls when trading rice, as noted by Radson: "I may use a bowl for selling broken rice, but not rice. We do not use buckets to ascertain the volume of rice." The men traders were very much in support of weight measures, although they may use volume measures.
While the majority of businesses in Kyela seem to fall within the micro and small business category in Tanzania (United Republic of Tanzania, 2002), we found women's businesses at Kalumbulu were constructed as relatively larger than men's businesses. This was evident from a man key informant, a processor: "Much of the rice in our storage facilities is owned by women. Women account for the majority with substantial capital." Our interview with Suzana, a woman who owns a processing facility and accesses bank loans for development purposes, further supported this. However, many other women traders at Kalumbulu ran relatively small businesses. These women were financially constrained. The men traders at Kikusya were found to be financially stronger, as farmers and supervisors at the facility showed men traders at the market to be reliable buyers. Women and other men at Kikusya operated small businesses and often sold rice on arrival to men rice agents stationed at the marketplace. The size of the business was often linked to the tools used to ascertain the quantity of rice, as can be seen from Gabriel, Jane's husband: "We men in rice, we play with bags. But the women are ready to trade with bowls [. . .]. . . . she starts with a bowl, then she moves to buckets, she finally finds herself in bags." The volume of business seems to matter to the men, whereas women can start small and grow over time. In the discussion of women's dominance in the rice business in Kyela, one woman key informant commented: "There are men trading rice, but their rice business involves large offices, packaging and selling rice from 20 kilograms." Women tend to be financially weak compared to men counterparts (Kuada, 2009). Thus, the study participant's claim that women traders at Kalumbulu are financially stronger than men seems to partly challenge the norm, while the claim that men traders at Kikusya are financially stronger than women is consistent with the norm.
We also found gendered symbolism in relation to professionalism and the branding of rice. Historically, rice sold in Kyela to both commercial customers and end consumers was unbranded. Recognising this as a weakness of the rice trading system, a few men traders saw the value of branding rice for trading reasons. Two of them, Peter, the owner of rice brand Kyela Rice, and Lusajo, the owner of Five Star Rice, have managed to establish businesses where they sell graded rice in labelled packages. Peter therefore claims to have introduced professionalism into the rice business in the district, and is in turn "self-promoting" (cf. Kelan, 2018). Indeed, some of the district policy makers cited Peter and Lusajo as role models. Their view of introducing professionalism seemed to give these men an edge in a sector that is constructed as feminine, similar to men childcare workers in Switzerland who secured their space in the sector by introducing new pedagogical approaches (Nentwich et al., 2013).
Flora, a woman rice trader at Kikusya and a widow, also saw the value of branding. She imagined the difference branding would make: "It is efficient to have rice in special packaging material with labels. That is different from the ordinary system where you just serve the customers. I mean, I use the packaging materials for the label. The packaging materials include your number [contacts]." Flora believed this would render trading efficient and improve professionalism. Despite her wish, Flora, who takes care of the family on her own, has not been able to implement her idea because of financial difficulties. Thus, while branding is a new rice business orientation, more men than women seem to be able to exploit the opportunities arising because they have better access to resources including knowledge and finance.

Interactions
With regard to interactions, we found practices of women and men distancing from each other. This was especially common among rice sellers, but we found differences between the two marketplaces. At Kikusya, the majority of rice traders were men, and rice business transactions took place within the processing facilities. Men, including supervisors, at this marketplace were negative towards women traders. They labelled women as too emotional in defending their continued use of buckets for measuring, or as in a hurry when serving customers, which Kelan (2018) calls ganging up on women. Men verbally discouraged women agents and brokers from entering the marketplace mainly to make the environment seem, what they referred to as, civilised. We found that men were distancing themselves from women in these ways for business reasons. A man agent at the marketplace said: "Ah! Here we have to be honest. We blind people. We restrict people, meaning we do not want rice brokers at this market." Also, during one of the FGDs, women traders at Kalumbulu marketplace described their blocked attempt to exploit the business opportunities at Kikusya marketplace.
At Kalumbulu, where the majority of rice agents, wholesalers and retailers were women, some of the women served customers from designated business premises. This was sometimes done jointly, in order to share costs, but it was also to support each other in serving customers. Women, however, avoided sharing business premises with men, in order to minimise social sanctions in the form of intimacy allegations. One woman trader, Rose, explained why women would not share business premises with men: "Imagine you are married. So cooperating with men, you know, there are always rumours. You will hear that the man she is sharing the premise with is her lover." While in this case women are distancing themselves from men, for fear of sexualisation of such relationships, Martin (2003) and Van Den Brink and Benschop (2014) reported men in developed countries distancing themselves from women for the same reasons. Limited possibilities for women in Tanzania to network with men has also been reported by Rutashobya et al. (2009). This may adversely affect women who depend on networks for raising finance for business, in addition to financial support from their male partners (Kuada, 2009).
Men also tended to control women's geographical mobility (cf. Porter, 2011), as we found women mainly trading rice at the marketplace while men were allowed to travel to the villages to buy rice. It is likely easier for men to monitor their wives at the marketplace than when women travel to villages to buy unprocessed rice. One key informant noted that market participants know each other and easily identify if someone is considered to be misbehaving. This implies the consistent monitoring of participants' behaviour at the marketplace, something that has also been reported from Kenya and Nigeria (Bergman Lodin et al., 2019). This is despite the fact that Kalumbulu is geographically wide, with multiple entries and diverse players involved in formal and informal rice business. Further, interaction between the women agents and retailers, and men buyers and brokers, was sometimes constructed at Kalumbulu as unpleasant for couples where both partners traded at the marketplace. One government officer, who is usually at Kalumbulu, also noted that some men interacting with women sellers seldom "use sexual language that may offend a women trader's husband who witnesses the conversation". We find that these interactions reproduce gendered divisions of labour and location in space; that is, men collecting rice from villages and women staying put at the marketplace.
The analysis also revealed practices of care in rice business. The importance of care in business is also reported in other sectors and countries (Anderson, 2008;Hall et al., 2007;McDonald, 2013). Women are normatively constructed as caring (Deutsch, 2007). Women sellers consistently cared for buyers, especially at Kalumbulu, where they are allowed and where the competition is strong. For example, retailers were quick to approach buyers as they arrived. They were friendly and welcoming, and used the argument of quality and a good price to persuade the customers to buy from them. One woman trader said: "A man cannot compete with us [women], because we are nice to customers. We are sharp and welcoming in business." The quote shows that women are proud of the caring attribute and capitalise on it to seek to outperform men. This aspect of gendered interactions thus seems beneficial to women, and may be a "special contribution" in business (Broadbridge & Simpson, 2011).
We found that some men were unhappy with the way women traders interacted with customers. Discussing the way women and men approach customers, Radson, a man rice agent at Kikusya, argued " [men] are not always in a hurry like our colleagues [women]". While the quote may denote unity between women and men through the phrase "our colleagues", it also signals differences regarding the way women and men interact with customers. Women tend to rush for customers as they enter the marketplace, whereas men stay calm while waiting for incoming customers. However, such women were not equated with "prostitutes", as reported from neighbouring Zambia (Evans, 2014). Still, men seem to dislike the women's approach, hence this becomes one of the reasons for men, the dominant players at Kikusya, to discourage women traders at this marketplace. Martin (2003) also found men underrating women's work and applauding men's work, regardless of its quality.

Conclusion
The aim of this paper was to understand and examine the ongoing developments and transformations in the rice sector in rural Tanzania from a gender perspective. Through focusing on two marketplaces in Kyela, Tanzania, we adapted the doing gender framework to explore how gender interactions are maintained and/or challenged in and through these transformations in terms of gendered identities, divisions, symbols and images, and interactions . We argue that gender is done by maintaining the norms about hegemonic masculinities and femininities, and that gender is undone by challenging these norms.
Our findings reveal doing gender practices to be more common than undoing gender practices among women and men in rice business and marketplaces. This implies that much more effort is needed to achieve gender equality, in spite of the progress noted. Further, more women than men were found to undo gender. It seems easier for women to act or perform in ways that are scripted as masculine than for men to perform gender in ways that are scripted as feminine. This could partly be linked to the fact that men and masculinity are constructed as superior to women and femininity (Acker, 1990;Pettersson & Cassel, 2014). Men tend to be cautious on what aspect of gender to undo, and deliberately avoid challenging their hierarchical superiority (Pierotti et al., 2018). Challenging the gender hierarchies seems problematic within the household but less so in the public space, as we show in this paper e.g., as men traders at Kalumbulu marketplace are accountable to women leaders.
With regard to gendered divisions, the traditional gender roles and work-home divide were challenged by these women. Unlike men who did not perform household care roles, women rice traders were found to provide for their families at the same time as they were dedicated to their businesses. The informal nature of the sector and electricity blackout in the district allowed, or indeed required, some women to work up to midnight, while others took their children to the marketplace. These practices are uncommon, or are rarely documented when studied, in corporate organizations, which often have clearly defined work rules, including arrival and departure times (Acker, 1990;Anderson, 2008). Unlike in some developed countries, where childcare facilities are well-developed and affordable (Pettersson & Cassel, 2014), similar facilities were unavailable in this context. The situation was compounded by men who were reluctant to handle housework and care responsibilities, thus reinforcing the norm and inequalities.
The norms limiting women's mobility were maintained, leaving benefits associated with movement to men (Mandel, 2004). Women were often stationed at the marketplace close to their homes, which entailed limited movements. Further, men stationed at the marketplace represented a challenge to the masculine norms granting them freedom of spatial mobility; however, these men focused on lucrative tasks such as brokerage at the old marketplace or trading graded rice at the new marketplace. Financial incentives seem to off-set the risk of men jeopardising masculinity (Nentwich et al., 2013). Men challenge norms of masculine mobility, but maintain the norm of controlling high paying occupations and products. Thus, not only are doing gender processes related in practice, as theorised by Acker (1990), but doing gender processes may also relate to undoing gender processes.
The ongoing transformation of rice from food to cash crop reinforces the image of men as financially and physically strong, role models, "rough", skilled in technology and professional. The constructions of professionalism in the rice business e.g., through the branding of rice, echo a debatable claim that men's entry into women-dominated sectors is associated with a rising standard in the sector (Nentwich et al., 2013;Tennhoff et al., 2015). Men are also taking over and crowding out women as the market value of rice increases (Doss, 2002;Njuki et al., 2011). Further, the establishment of the new marketplace seems to reverse the widely held narrative that women in Kyela are financially stronger than men. Our findings revealed men to be financially stronger traders than women at the new marketplace. Moreover, women had limited access to weight measuring technology, which further demonstrates how access to assets and resources is gendered. Structural and material inequalities also contribute to the doing of gender (Pierotti et al., 2018), but are insufficient to explain persistent gender inequalities (Deutsch, 2007).
While Martin (2003) argues that men unconsciously mobilise masculinities, the men at the new market consciously favour fellow men because they are viewed as more skilled than women. As expected, women consciously avoided interacting with men in order to minimise possible intimacy allegations which could ruin marriages (Carr & Kelan, 2016;Martin, 2003;Rutashobya et al., 2009). Also, the traders at the new marketplace seem to underrate the value of customer care in the rice business, possibly because of a readily available market. However, women at the old marketplace capitalise on customer care, especially due to high competition at this marketplace. Thus, care being constructed as feminine gave women a competitive advantage in the rice business (Broadbridge & Simpson, 2011). Notably, despite the transformation in the sector, conventional gendered interactions are maintained.
As can be deduced, the structures of the marketplaces-their organizational logic (Acker, 1990(Acker, , 1992)-seem to influence the doing and undoing of gender. For example, the old marketplace (Kalumbulu) offers more avenues to challenge gender interactions, whereas the new marketplace (Kikusya) mainly reinforces dominant gender interactions. Women at the new marketplace (Kikusya) consciously or unconsciously perform lower paid and more flexible tasks, similar to practices in formal settings (Anderson, 2008;Gherardi & Poggio, 2001). The old marketplace (Kalumbulu) is led by women and permits entry of both women and men traders, thereby partly lessening the gender gap. However, men still handle the brokerage role at that marketplace, which gives them a higher income than women traders (dealers and retailers). Men, who are less prowomen traders, lead the new marketplace (Kikusya).
At face value, the adoption of high-tech processing facilities, currently at the infant stage but spreading, seems to benefit the rice sector and Kyela community by encouraging value addition to rice and taking the markets closer to farmers as set out in national strategic documents (United Republic of Tanzania, 2016, 2019). Khoza et al. (2019) reported similar findings with respect to proximity to agro-processing in South Africa, but remained silent on gender. However, there are drawbacks associated with the introduction of the new marketplace in Kyela and its new technology. First, as expected, men are reported to be controlling the trading of rice at new and/or lucrative markets (Andersson et al., 2016;Kantor et al., 2015;Mudege et al., 2015), which has also turned out to be the case in Kyela. Unfortunately, the strategic location of the new marketplace and the nature of rice traded at this market seem to further reinforce gender inequalities. The new marketplace trades graded rice and is located along the main road used by most buyers as they enter Kyela town centre, where the old marketplace is located. This seems to be causing a shift of the major commercial buyers from the old to the new marketplace, disempowering women, who are the majority of players at the old marketplace. Second, technology in itself limits women's access to the market, as technology is often linked to men and masculinity (Kantor et al., 2015). In particular, while the intention to establish Kikusya marketplace and introduce new technology aimed at achieving inclusive participation, the current set-up reinforces gendered divisions in favour of men. This is not a surprise as technocrats are rarely gender savvy and seldom take a transformative approach to gender equality (Resurrección & Elmhirst, 2021). Third, the fencing of the government processing facility at the new market, possibly for safety and security reasons, seems to implicitly symbolise privacy. This also seems to favour men, who tend to be entrusted to control resources by the norms or regulatory institutions (Amine & Staub, 2009;Quisumbing & Pandolfelli, 2010).
In sum, the transformation has resulted in more processes of doing gender than of undoing gender, which is subsequently reinforcing the prevailing gender inequalities, especially in developing countries and more specifically in rural Tanzania. As postulated by Doss (2002) and Njuki et al. (2011), women rice traders are increasingly being replaced by men as rice transforms from a food crop to a cash crop. Over time, this may reduce women's control over rice income, as the crop falls in the hands of men (cf. Aberman & Roopnaraine, 2020). This is alarming given the fact that women spend most of their income on household food, health and education (Mandel, 2004), while men do not.
Interventions that do not give equal weight to the technical and social aspects are likely to yield unexpected outcomes (Amine & Staub, 2009;Kantor et al., 2015). Further, interventions, also those departing from good intentions, do not automatically translate into equitable benefits for women and men. Men may value their hierarchical superiority in the community thus deliberately block attempts to create a more egalitarian community (Pierotti et al., 2018). It is through continuous commitment over time and of other resources as well as guided implementation that interventions can yield the expected outcomes (Bajaj, 2009;Kantor et al., 2015;Njuki, 2016). This is critical in economically important and dynamic sectors such as rice in Tanzania, part of sub-Saharan Africa, characterised by attitudes, beliefs and norms that often work against women (Amine & Staub, 2009;Rutashobya, 1998). Future agricultural projects in this context, including those for scaling-up technology and marketplaces, should therefore apply gender transformative approaches (GTA) (Kantor et al., 2015;Njuki, 2016). These involve exploring underlying causes of gender inequalities and give equal emphasis to technical and social issues in interventions. As such, GTA are more likely to produce optimal socio-economic results. Incorporating GTA in future interventions is critical as technocrats including gender advisors or "gender experts" (e.g., community development officers responsible for gender mainstreaming in Tanzania) often focus on simplification and reporting on gender, while losing the transformative agenda (Resurrección & Elmhirst, 2021).
Our paper makes four major contributions. First, it adapted Acker's doing gender framework by combining doing and undoing gender perspectives. We applied the developed tool to explore the everyday practices of women and men in the business and marketplaces for rice, a traditional food crop controlled by women, which is transforming into a cash crop. The tool has allowed us to uncover some taken-for-granted practices on gendered labour divisions, symbols, interactions and identities, which often work against women. Second, the paper not only supports the claim that doing gender processes in practice are related (Acker, 1990), but it further argues that so are undoing gender processes. Third, the paper is one of the few comprehensive studies in sub-Saharan Africa applying Acker's doing gender framework to study how gender is reinforced and/ or challenged in the same gender system. Fourth, through the analysis of the two marketplaces, the paper demonstrates the way structures of workplaces facilitate the doing or undoing of gender. In practice, the stakeholders need to promote structures that facilitate undoing gender practices, as they contribute to gender equality.
Future studies may apply the updated doing gender framework in examining gender in other rice, food or related informal markets. This may include use of an alternative approach to undoing gender (Butler, 2004), which focuses on discourse analysis. Also, much of the empirical material for this paper is based on interviews, which are, unfortunately, less effective in capturing the saying and doing of gender (Martin, 2003). Ethnographic studies with extensive fieldwork may address this limitation.