The study was conducted in Benin among agricultural advisors. The choice of this target group is justified by the position of agricultural advisors at the interface between the technical orientations of agricultural development programs and the socio-economic, environmental, and cultural realities of farms. This position exposes them to situations in which several professional norms, objectives, or constraints may come into tension, particularly in the context of the agroecological transition.

Participants were selected using purposive sampling. Selection prioritized advisors working with producers within accredited service-providing structures, notably farmer professional organizations (FPOs), non-governmental organizations (NGOs), and private consulting firms. Participation was also based on advisors’ consent and their availability at the time of data collection. In total, 30 agricultural advisors took part in the study. The diversity of professional profiles was taken into account in order to gather varied positions regarding the tension situations advisors may face in the exercise of their duties. Table 1 presents the profile of these advisors.

Table 1. Profile of the surveyed advisors.

Individual interviews were conducted with advisors to gather information using an interview guide. To be selected to participate in the study, advisors had to be affiliated with an accredited service provider (Table 1), work as agricultural advisors in the field with producers, and have at least two years of experience in this role. This selection was carried out with the authorization of the head of the advisory team at each organization, who was first contacted for a presentation of the study; thereafter, advisors meeting the eligibility criteria were invited to participate on a voluntary basis and subject to their availability. The interviews were conducted in person and in French (the common working language for both the interviewer and the participants) at a location and time chosen by the advisor based on their geographic location.

All interviews were recorded with the participants’ consent and transcribed in full in French shortly thereafter. They were conducted by the authors, who had no hierarchical or institutional ties to the participating advisors or their employing organizations, thereby minimizing bias in responses driven by social desirability.

The data collected related to the sociodemographic characteristics and career paths of the advisors, the methods and practices used in providing advice, perceptions of the role of the advisor in conventional agriculture and agroecology, as well as the positions and stances adopted in response to professional dilemmas. In total, five dilemmas were presented to the participants for their assessment. For each situation, they were asked to indicate the approach they would favor and to justify their choice. For each dilemma, participants had to identify the single approach they were most likely to favor. When an advisor’s response combined elements of both approaches, the response was classified according to the approach corresponding to the final choice expressed.

Construction of the Professional Dilemmas

The professional dilemmas were constructed on the basis of the roles of the agricultural advisor as defined by the [14] in the context of the agroecological transition (Table 2). Each dilemma opposes two analytical orientations: one associated with the principles and modalities of conventional agricultural advisory work, and the other with an agroecological logic. These professional dilemmas are designed to test advisors’ stated representations of their role and their choices when faced with concrete professional situations.

Table 2. Summary of the expected roles and responsibilities of the agricultural advisor in agroecology.

Source: adapted from [14].

Dilemma 1: Tension between the application of a standardized technical recommendation and its adaptation to local realities. It confronts the advisor with a situation in which a national recommendation or project protocol must be applied in a context where its implementation modalities may be difficult to reconcile with the characteristics of the environment. The advisor must therefore arbitrate between complying with the technical prescription and adapting it to the realities of the intervention environment, at producers’ request.

Dilemma 2: Conflict between compliance with specifications excluding the use of synthetic chemical products and a producer’s decision to resort to synthetic chemical pesticides following a pest attack likely to severely affect production and income. This situation confronts the advisor with a trade-off between complying with the environmental requirements of the intervention program and seeking an immediate response to an economic and productive difficulty.

Dilemma 3: Opposition between an approach based on participatory learning and the co-construction of knowledge, through Farmer Field Schools (FFS), and an approach favoring the direct transmission of technical recommendations enabling rapid adoption of practices. This tension is heightened by the time constraints faced by producers and by the pressure exerted by the project to achieve its adoption targets within the allotted time. The dilemma thus makes it possible to examine how advisors arbitrate between the time required for a collective, participatory learning process and the imperative of rapidly achieving measurable changes in practice.

Dilemma 4: The advisor’s scale of intervention when faced with an agricultural problem whose causes extend beyond the farm. It sets in tension an intervention centered on the producers’ farms and an approach based on mediation and coordination among the various actors of the territory. The situation confronts the advisor with the need to take into account factors such as land pressure, the management of livestock grazing routes, and access to biomass, while considering the existing tensions between farmers and herders. The dilemma makes it possible to examine how the advisor arbitrates between an intervention limited to the concerns of the producers he or she supports and involvement in the concerted search for solutions at the territorial scale.

Dilemma 5: Women’s participation in advisory activities and community-level decision-making processes. It confronts the advisor with a situation in which the requirements of inclusion and participation of women may come into tension with certain social norms and traditional practices of the community. The advisor must then arbitrate between taking into account local norms likely to condition the acceptability of the intervention and the need to foster the effective participation of women.

The transcribed interviews were cleaned up before being analyzed. Two complementary approaches were used with the IRaMuTeQ 0.8.4 and NVivo 14 software programs. Using IRaMuTeQ, a lexical discourse analysis was conducted to identify and compare the main terms and semantic categories associated with the role of the advisor in conventional agriculture and agroecology. The corpus included all responses relating to the advisor’s role in each model (287 distinct lexical forms for the conventional agriculture corpus and 544 for the agroecology corpus, prior to lemmatization), with each response treated by IRaMuTeQ as a distinct textual unit. The lexical forms were lemmatized using the lemmatization tool integrated into IRaMuTeQ, and a standard list of stop words (articles, prepositions, pronouns, and auxiliary verbs) was applied before calculating frequencies and generating word clouds. The terms selected for the presentation were then translated from French into English by one of the authors. These translations were subsequently verified against the original transcripts to ensure that the contextual meaning was preserved. With NVivo, the five dilemmas were subjected to content analysis to identify, for each dimension, the preferred orientations and the arguments supporting them. The responses to the different dilemmas were then compared with the discourse of the same advisors to examine convergences and variations in their positioning.

The six semantic categories used for the quantitative comparison (posture and intervention logic; object of advisory work; knowledge transmission and construction; human support and empowerment; economic and productive outcomes; environment, sustainability, and health) were defined inductively from an initial reading of the transcripts and then refined through discussions between the two authors. The coding unit was the meaningful text segment (a proposition or sentence expressing a single idea concerning the advisor’s role), rather than the interview as a whole or an isolated lexical form. The categories were considered mutually exclusive; that is, when a segment fell under several dimensions, it was assigned to the category reflecting its dominant content. The authors coded the entire corpus and compared the category assignments. Any disagreement was resolved through discussion until consensus was reached.

Participation in the study was based on advisors’ free consent, obtained at the beginning of each interview. Participants were informed of the study’s objective and of the scientific use of the data collected. The anonymity and confidentiality of the information provided were preserved during processing, analysis, and presentation of the results. The statements reported in the article do not allow individual participants or their institutional affiliation to be identified.

Figure 1 and Figure 2 present the word clouds derived from advisors’ discourse concerning their respective representations of the advisor’s role in conventional agriculture and in agroecology. The underlying lexical analysis was conducted on the original French-language interview transcripts using IRaMuTeQ; each term has been translated into English here, with word size preserved in proportion to its original frequency/prominence in the French-language corpus.

The lexical field of the agricultural advisor’s role in conventional agriculture comprises a total of 287 distinct words. It emerges that respondents’ perceived role of an agricultural advisor in conventional production is organized around three categories of variables.

Figure 1. Lexicon reflecting the advisor’s role in conventional agriculture.

Figure 2. Lexicon reflecting the advisor’s role in agroecology.

The first groups words such as chemical, pesticide, synthesis, producer, fertilizer, input, conventional. The advisor thus appears as an actor responsible for directing the producer toward the use of synthetic chemical fertilizers and pesticides. The second, relating to the quantitative performance of agricultural production, is organized around yield, physical, result, maximize. The advisor is here understood as an actor whose objective is to improve production by increasing yield. The term maximize suggests that agricultural performance is regarded as a function of the quantity produced. Synthetic inputs are thus not represented as an end in themselves but rather as the technical means mobilized to achieve an increase in the producer’s yield. The last category reflects the normative framework the advisor follows, and includes words such as conventional, model, systematic, standard, rigid. These terms suggest that the support provided by the advisor follows relatively stabilized norms and procedures.

Unlike the lexical field of the advisor’s role in conventional agriculture, that of the advisor in agroecology comprises 544 distinct words, almost double the conventional corpus. Analysis of the word cloud shows that agricultural advisors apprehend their role in agroecology through concepts that are likewise organized around three categories of variables. The first relates to natural resources such as soil, land, crops, plant, tree. The advisor is regarded as an actor whose support is based on living components and the natural resources that can be mobilized in the production process.

The second category, relating to pedagogy, is organized around the words teaches, learns, method, technique. The advisor is here understood as an actor who continues to bring and transmit innovations and techniques to producers, as suggested notably by the presence of the term teaches. However, the advisor’s representation is not limited to a prescriptive, top-down posture. The presence of the term learns also suggests that the advisor is himself or herself placed in a dynamic of learning through contact with producers. The advisor thus appears as an actor who accompanies the producer in acquiring knowledge and techniques, while also being liable to learn from the producer’s experience and practices. This configuration therefore points to a function oriented more toward mutual learning and the co-construction of knowledge.

The last category, relating to local anchoring, groups words such as local, own, naturally, resource. These terms suggest that the advisor’s support takes into account the specific characteristics of the production context and the resources available locally. The word chemical, which appears among the terms used, does so in a logic of reduction or avoidance (reducing the use of chemical products, or helping to avoid them), which contrasts with its place in the lexical field of the role in conventional agriculture.

In sum, in conventional agriculture, the advisor is regarded as an actor who introduces a practice or innovation by following technical norms that encourage the use of synthetic chemical inputs with a view to increasing the yield of the producer being supported. By contrast, unlike the conventional model, where synthetic inputs appear as technical means associated with yield maximization, the representation of the advisor’s role in agroecology appears to be organized more around the consideration of natural resources, learning, and adaptation to the local context.

Figure 3 presents the semantic distribution of the roles perceived by advisors in conventional agriculture and in agroecology.

The semantic categories most represented in the conventional model are posture and intervention logic (30.84%), the object of advisory work (24.27%), economic and productive outcomes (15.27%), and knowledge transmission and co-construction (14.96%) (Figure 3). Conversely, in the representation of advisory work in agroecology, the most important categories are environmental protection, sustainability and soil health (23.2%), knowledge transmission and construction (22.58%), the object of advisory work (21.14%), and posture and intervention logic (16.82%).

These results reinforce the previous findings and highlight a shift in the representation of the advisor’s role. Whereas advisory work in conventional agriculture appears to be structured mainly around a prescriptive logic oriented toward increasing yield through the use of synthetic chemical inputs and compliance with standardized norms, the representation of advisory work in agroecology gives greater weight to environmental protection, sustainability, soil health, and processes of knowledge construction. This shift is particularly visible in the decline of the posture and intervention logic category, which falls from 30.84% to 16.82%, and in the rise of knowledge transmission and construction, which increases from 14.96% to 22.58%. The environment, sustainability and health dimension also shows a significant increase, from 10.38% to 23.2%. The advisor thus appears to be represented as less centered on a predefined intervention logic and more oriented toward learning, knowledge construction, and consideration of the ecological dimensions of production. Nevertheless, compliance with standardized norms remains present in this representation.

Figure 3. Semantic distribution of roles perceived by advisors in conventional agriculture and in agroecology.

The human support and empowerment category, although less represented, also shows an increase, from 4.27% in conventional agriculture to 7.41% in agroecology. This shift points toward a representation of advisory work more closely associated with producer empowerment, notably through the valorization of the producer’s agricultural assets and the pursuit of their profitability, rather than through the sole transmission of technical solutions. Furthermore, economic and productive outcomes decline from 15.27% in the conventional model to 8.85% in agroecology. This decline does not mean that economic performance becomes less important in agroecology, but rather that it occupies a relatively smaller place in the discourse associated with the advisor’s role.

In sum, the word clouds provide an immediate and intuitive reading of the lexical contrast, while the semantic categorization allows this contrast to be objectified and quantified. The two analyses reinforce one another in showing that conventional advisory work is characterized more by its method of intervention and its productive outcomes, whereas agroecological advisory work is characterized more by its relationship to living systems, to learning, and to the construction of knowledge.

Fisher’s exact test confirms that representations of the agricultural advisor’s role differ significantly between conventional agriculture and agroecology for five of the six categories studied (Table 3). These differences are significant for posture and intervention logic, knowledge transmission and construction, human support and empowerment, economic and productive outcomes, and environment, sustainability and health. The differences observed in these dimensions therefore do not merely reflect descriptive variations in the corpus but correspond to statistically significant differences between the two representations. By contrast, no statistically significant difference is observed for the object of advisory work (inputs, actors, resources). Despite the general reconfiguration of the advisor’s perceived role between the two models, the elements relating to the object of advisory work thus remain relatively stable.

Table 3. Significance test of the perceived differences between representations of the advisor’s role in conventional agriculture and agroecology.

*Significant at the 1% threshold. AC: Conventional Agriculture; AE: Agroecology.

Table 3 presents the exact Fisher’s tests performed on individual word occurrences classified in the two corpora, totaling 655 occurrences for the conventional corpus and 1457 for the agroecological corpus. The unit of analysis is the lexical occurrence, not the advisor. The tests thus compare the relative frequency of each category between the two corpora. However, since all 30 advisors contributed to both corpora, the occurrences are not entirely independent in the strict statistical sense. The results are therefore interpreted at a descriptive level, as indicating differences in lexical composition between the two discourses, rather than as individual differences among advisors.

The analysis reveals significant differences (p < 0.001) for five categories: intervention approach and logic, knowledge transfer and construction, human support and empowerment, economic and productive outcomes, as well as the environment, sustainability, and health. These results indicate a differentiation in the lexical composition of conventional and agroecological representations across these dimensions. In contrast, no statistically significant differences were observed regarding the focus of advisory work (inputs, actors, resources; p = 0.113), suggesting relative stability in this dimension between the two representations.

Figure 4 presents the orientations expressed by advisors with respect to the five professional dilemmas.

Figure 4. Orientations expressed in response to the five professional dilemmas.

The analysis of each of the five dimensions of the advisor’s expected role reveals a dominant posture, which the following verbatim excerpts illustrate.

Regarding the frame of reference for action, contextualization of interventions clearly dominates compliance with standardized prescriptions (73% versus 27%). Advisors justify this posture by their proximity to producers and their knowledge of local realities, even if it means departing from institutional recommendations: “The project I work on recommended that we use campeche wood(Haematoxylum campechianum)for the live hedge.But for two years it would not grow.A producer then suggested we use jatropha instead,since it is available locally,and barely three months later the plants had really grown”. The opposite posture remains a minority position and stems more from contractual constraint than from technical conviction, as expressed by an advisor from UNCPS: “I do what the project wants me to do,since it is the one paying my salary that gives the orders”.

Regarding the purpose of advisory work, the dominant posture departs from strict compliance with the agroecological objective (23% versus 77%), condoning the producer’s recourse to synthetic chemical inputs in the face of the producer’s economic urgency, as explained by an advisor from AIC: “If the producer lost his crop,whatever you told him afterward,he would never listen to you again.It is best to accept his decision so the harvest can be saved”. This concession is also driven by the vulnerability of the advisor, who often lives within the community he or she supports. The minority posture, in this case the one aligned with the agroecological objective, instead relies on close support intended to ensure the proper application of biological solutions. As an advisor from CRADIB puts it: “The advisor has to follow the producer every day,at every step of the process”.

As for the mode of knowledge transmission, participatory learning dominates direct technical transmission (73% versus 27%), driven by the demonstration effect among peers: “It is when they have someone with a field next door who has implemented a practice,and it has worked,that is when they too will try to reproduce it” (advisor from UNPCS). Direct technical transmission, a minority position, remains a pragmatic response to producers’ low turnout at Farmer Field Schools. As an advisor from Sian’son states: “Out of a class of30learners,you can hardly find10”.

Regarding the scale of intervention, territorial mediation prevails over intervention limited to the farm (77% versus 23%), notably in farmer-herder conflicts: “The producer has a critical need for organic matter to sustainably regenerate his land.The herder also has an essential need to access crop residues” (advisor from MAP). The farm-centered posture, a minority position, mainly stems from caution in the face of physical risk. As an advisor from UNPCS puts it: “If the local authorities themselves cannot resolve the problem,it is not up to the advisor to do so”.

Finally, regarding principles of action toward social norms, the inclusion of women dominates, through strategies of accommodation with local norms (87% versus 13%) rather than through their outright challenge: “In several communities in northern Benin,women are afraid to speak certain truths in front of their husbands” (advisor from UNPCS). The posture of compliance with socio-cultural norms, a minority position, does not necessarily abandon this objective but favors a gradual approach.

Thus, these five dilemmas confirm, on the basis of advisors’ individual responses, the same orientation suggested by the reading of the word clouds: four dimensions clearly favorable to the agroecological pole, with the purpose of advisory work remaining the most contested.

The analyses reveal a dichotomy of postures, one oriented toward conventional agriculture and the other toward agroecology (Figure 5).

Agricultural advisors whose representation is oriented toward conventional agriculture display fairly similar postures regarding the pursuit of productive and economic performance, equality/inclusion, and territorial mediation and coordination, regardless of years of experience.

Figure 5. Distribution of advisors by intervention posture and professional seniority (conventional agriculture and agroecology posture).

By contrast, the two groups of advisors whose representation is oriented toward the agroecological model show an almost identical posture, namely contextualization of interventions, participatory learning, and co-construction. A difference does appear, however, concerning compliance with agroecological principles, which is more present among advisors with fewer than five years of experience than among those with more than five years. Advisors with more than five years of experience also appear to place greater emphasis on farm-centered intervention.

Ultimately, seniority does not emerge as a major factor of differentiation in advisors’ perceived posture. Both groups share common orientations in both the conventional and the agroecological models, although the way they perceive the advisor’s role shows marked differences. The differences observed relate more to the intensity given to certain dimensions: more experienced advisors appear more oriented toward productive performance in the conventional model, whereas those with less experience show greater interest in agroecological principles. The most stable element in the agroecological representation nonetheless remains the contextualization of interventions and participatory, co-constructed learning, present in both groups.

This study presents several limitations. First, the sample size, limited to 30 advisors, does not allow the results to be statistically generalized to all agricultural advisors. Second, the professional dilemmas make it possible to grasp the postures advisors say they would favor when faced with given situations, but they do not make it possible to directly measure the practices advisors would adopt in real-world situations. The results therefore provide more information about declared representations and stated positions than about actually observed behavior.