How Cases Speak to One Another: Using Translation to Rethink Generalization in Political Science Research

Abstract

Regardless of method, political scientists often seek to develop arguments that can be generalized to a population of cases. But is this the only way to think about how cases speak to one another? We advocate for a new way to think about how qualitative research produces broadly applicable insights: translation. Much like linguistic translation, the goal of translation in political science is to develop ideas that are intelligible in a different context, even as the context will change how an idea or political practice is interpreted or enacted. Translation offers at least three benefits. It allows us to (1) rethink how we form and deploy concepts; (2) rethink what a generalizable argument is by carrying parts of an argument, instead of entire causal chains to other cases; and (3) rethink how we conceptualize knowledge accumulation to include an abductive process where generating theory is the primary goal.

What do political scientists mean when we say our arguments generalize? More specifically, do scholars from different epistemological communities mean the same thing when we claim an argument generalizes? If not, can there be more than one way to think about what generalization entails? If there are multiple ways to think about generalization, does the language of generalization effectively describe how claims derived from our primary cases speak to other cases? These fundamental methodological questions shape not only how we communicate with other scholars but also how we imagine the possibilities for our research, including the kinds of questions we ask, the concepts we deploy, the research designs we develop, and the ways we build on existing scholarship to understand politics.

To make general claims, political scientists typically deploy one of three approaches. The first identifies empirical regularities across a large population of cases (see esp. King, Keohane, and Verba 1994)—say, the relationship between wealth and the likelihood of democratization (Przeworski et al. 2000). The second approach pursues descriptive or conceptual abstraction (see esp. Sartori 1970)—say, a model of democracy as a set of rules for settling political conflicts without bloodshed (Bobbio 1991). Each of these modes of theorizing might be described as developing accounts of politics as if they were a view from nowhere, creating general descriptions of politics untethered from particular times or places (Przeworski and Teune 1970; see also Cheesman 2021). In contrast, a third approach seeks middle-range theory that is derived from particular cases and bounded by scope conditions (see esp. Merton 1996, chap. 3)—say, explaining how labor mobilization shaped democratic and authoritarian coalitions in Latin America (Collier and Collier 1991). ¹

Tensions, however, are built into and exist among these three approaches. In the first instance, scholars pursuing empirical generalization remain tethered to average effects and often cannot explain specific cases. In the second, scholars engaged in conceptual abstraction face difficult choices about the appropriate level of abstraction for an individual concept given the particularities of the specific cases they want their concepts to describe. They also face difficulties in capturing how concepts work in practice in the places they study. In the third, scholars developing middle-range theory face constraints in the breadth of their claims, an especially difficult challenge in a discipline that often prizes broad theories over narrow insights. Furthermore, when we look at all three approaches together, there are clear tensions in what political scientists mean when they talk about generalizability. Each model pursues a different notion of generalizability—from average effects, to conceptual breadth, to causal accounts that identify specific mechanisms at work within a delimited range. This suggests that scholars often speak past each other when they talk about generalizability.

Responding to these tensions, we develop a fourth approach to generalization. As we argue below, the approach offers a view from somewhere that develops insights on politics with broad relevance (Anderson 1998; Cheesman 2021). In contrast to practices of empirical generalization and conceptual abstraction, this approach assumes that a view of the whole is always contingent, partial, and situated. In contrast to middle range theory, it assumes that not every part of an argument needs to carry beyond the study’s scope conditions for the theory to have broad relevance. Thus, our approach is simultaneously situated in specific empirical contexts even as it seeks to speak to cases outside of specific scope conditions. We call this approach to generalization translation. ²

To illustrate the value of this approach, we proceed as follows. First, we consider the limitations of current approaches to generalization for qualitative case-based research. Second, we describe how the practice of translation can help cases speak to one another. Third, we detail the epistemological assumptions underlying translation by examining how professional translators approach their work. Fourth, we apply this approach to political research, showing how qualitative political scientists are often translating without conceiving of it as such. We also show how explicit use of the translation language would allow scholars to describe their contributions more effectively while broadening the available goals of their research. We conclude by discussing how we might judge the quality of a translation. To start, though, we consider the challenges of existing approaches to generalization for qualitative, case-based research.

EXISTING LOGICS OF GENERALIZATION AS A PROBLEM FOR CASE-BASED RESEARCH

Why do we need a new way to think about how cases speak to one another? A first issue is epistemological. Even as there have long been attempts to develop unified logics of social scientific explanation, scholars from different epistemological communities have fundamentally different approaches to explanation. This makes the languages of statistical generalization or conceptual abstraction poor fits for qualitative research (Small 2009), ³ creating a dilemma for case study researchers. As Seawright and Gerring (2008, 294) argue, because most case studies seek to say something about a broader population, “the chosen case is asked to perform a heroic role: to stand for (represent) a population of cases that is often much larger than the case itself.” However, this inherent heroism places case study researchers in a bind such that broader claims issued from specific cases may stand on shaky ground.

Responding to this dilemma, scholars have developed a battery of case selection strategies—including the selection of extreme, deviant, typical, most likely cases, and more—to fortify their ability to make broad inferences. However, each strategy comes with serious costs, including (among other challenges) the possibility of distorting a case to fit into a prior theory, overclaiming on the basis of the best-known cases, losing representativeness when examining extreme cases, and many other challenges. ⁴

And there is a deeper problem. Trying to make generalizable claims from a specific set of cases assumes ex ante that one knows the set of cases to which a set of arguments should apply (Gerring 2004). ⁵ This presents dilemmas for scholars engaged in theory development. It necessarily rules out the possibility of a theory speaking to cases outside of the intended sample and problematically assumes the scholar knows the relevant set at the beginning of the research when, in fact, the constitution of a set is often an outcome of the research process (Soss 2021). ⁶ So, case study researchers are placed in both an ontological and a methodological bind when they seek to make broad theoretical claims from a small set of cases.

Deep knowledge of cases, combined with a hesitancy to see cases as representative, can, understandably, lead qualitative scholars to be cautious when they talk about how their cases speak to others. This is particularly true when researchers identify the specific mechanisms that generate a particular outcome. Motivated by an effort to ensure that all the causal processes at work in the cases they study are replicated in the cases to which they think their argument speaks, many case study researchers set tight scope conditions around their theory; failing to do so could undermine confidence that the mechanism has been properly described (Mahoney and Goertz 2006, 237–8). Consequently, to the degree that case study researchers working without a large, representative sample make general claims, they often only do so at the middle range (Ahram, Köllner, and Sil 2018, 15–6; George and Bennett 2005, 266)—a choice that inherently limits the potential utility of a given study because it necessarily reduces the scope of a theory. ⁷

These dilemmas, however, need not relegate case study researchers to providing descriptive accounts appealing largely to specialists or to setting tight scope conditions on when and where their theory might apply. Rather, we see them as an invitation to rethink how our cases speak to one another in the first place. The language of translation, we argue, helps us do precisely that by offering both a powerful addition to existing logics of generalization and alternative approaches to the process of research itself.

TRANSLATION AS A WAY CASES SPEAK TO ONE ANOTHER

We define translation as a recursive process of making sense of ideas or phenomena across two or more contexts with the goal of illuminating family resemblances in the concepts, political practices, or causal processes among them. Much like textual translation, the practice of translation in political science assumes that ideas, practices, or processes are comprehensible across different contexts. Importantly, however, they may not work in precisely the same manner as the context changes. A word in translation, for instance, may not carry the same meaning as its closest equivalent in another language, but the meaning is typically still understandable and interpretable, which can facilitate communication and understanding of both similarities and divergences. We argue that scholars can think about political practices in translation in the same way.

As we describe below, linguistic translators navigate imperfect options for how to convey meaning across languages, as certain words or phrases lack a perfect equivalent in the destination language. To respond to this dilemma, translators adopt a pragmatic approach. They make choices about the best word or phrase to use, and, through a combination of trial, error, experience, and creativity, they make these choices work. The same is true, we suggest, about how social scientists understand their cases and how they think about those cases in relation to others. Frequently, they extend theories to cases in the absence of clear rules of inference through a similar, if often unstated, creative pragmatism, drawing connections that illuminate similarities across space or time that are not bounded by strict scope conditions. ⁸ As Clifford Geertz (1973, 23) argued of our inferential choices generally, “where an interpretation comes from does not determine where it can be impelled to go.” Geertz’s insight extends to the ways in which we develop theory: cases speak to one another because they are made to.

Thinking about our ability to creatively make cases speak to one another frees us from the bind of seeing the goals of our research as the search for empirical regularities, average effects, and broad conceptual abstractions or having to restrict our causal and theoretical claims to the middle range. Through translation, we can simultaneously ground ourselves in local empirical patterns, theorize broadly about how those patterns work elsewhere, and think abstractly about how those patterns are conceptualized from the situated contexts in which we work. The approach uses local ways of seeing the world to look outward on politics without the bounds of strict scope conditions. It also frees us from needing an entire causal chain to work across cases, so that the logic of testing, replicating, and potentially falsifying an argument may not apply to work that seeks to translate, as political scientists would not assume that an entire causal process works in exactly the same way elsewhere. Further, when we think about political practices as translating, we do not assume that the set of cases to which an argument might apply can be known beforehand, nor that the scholar’s first understanding of the possible sets to which an argument may apply is the best casing for that argument (on casing, see Soss 2021; see also Becker 1998).

In this way, by conceiving of our arguments as translating, qualitative scholars will not only be able to more effectively describe how our cases speak to others but also expand the possibilities for theoretical innovation. ⁹ Specifically, we contend that thinking of our arguments as translating has at least three distinct advantages for qualitative scholars. It allows us to (1) rethink how we form and deploy concepts; (2) rethink what a generalizable argument is by carrying parts of an argument, instead of entire causal chains to other cases; and (3) rethink how we conceptualize knowledge accumulation from a process of empirical observation, inductive hypothesizing, and deductive testing to also include a recursive, abductive process where generating new theory is the primary goal.

Importantly, some political scientists already engage in the practice of translation—but they do not name it as such. ¹⁰ The costs of this conceptual deficit are severe. When scholars do not have the language to describe how their claims speak to additional cases (or even how cases within their study relate to one another), their arguments are necessarily truncated because they lack the ability to describe how part of an argument, one specific causal mechanism, or one element of a concept, carries beyond the original cases. Consequently, scholars face a bind: either they overclaim how their cases speak to others because they want their seemingly narrow study to be theoretically relevant or they underclaim by saying they cannot speak to cases beyond those within their scope conditions where their arguments are strongest (Gerring 2017, 222). The language of translation would allow scholars to get out of this false bind by opening the ways in which cases might speak to one another, even if imperfectly.

Moreover, for those scholars who do employ approaches like those we describe as translation, they typically do so in an ad hoc or informal manner. This limits the ability of other scholars to use those logics themselves and to effectively build on others’ theoretical claims and empirical insights. By describing these ad hoc practices as a process of translation, we seek to bring ontology and methodology together (Hall 2003). That is, we seek to have the discipline preach what it already practices—an outcome that will enable scholars (and particularly junior scholars) to more effectively describe how their cases speak to other times and places and think more creatively about how to build on the insights of others even as they develop their own. This will be particularly true if the language of generalization as currently conceived in the discipline is a poor fit for their research. ¹¹

Therefore, by formally naming these existing informal practices as acts of translation, scholars can systematically describe how their conceptual contributions might speak to many times and places, show how their theoretical contributions shed light on cases outside the bounds of strict scope conditions, even if entire theories don’t travel, and reconceive how their work builds on earlier generations of scholarship. Taken together, these contributions allow scholars to see their own work and how it relates to other theories, times, and places differently. Scholars will ask different questions and see new and different connections. As a result, they will develop novel theoretical understandings of the world that would have not been possible had they felt bound by existing logics of generalization.

In what follows, we focus on how the practice of translation might be useful for qualitative research: our own training as qualitative scholars means these are the logics we know best. We hope, however, that conceptualizing translation will be useful to scholars regardless of the methods they use. But we defer to scholars versed in other approaches to determine that. We also acknowledge that translation may not make sense for every project. Our goals, therefore, are modest: (1) to outline how the practice of translation offers useful approaches through which scholars can see how their cases might speak to others and (2) to show how translation offers an effective way of describing how scholars make their cases speak to one another. To achieve these goals, we need to understand the epistemology that underlies translation. To do so, we explore how professional translators approach their craft—a goal to which we turn next.

THE CRAFT OF TRANSLATION

At its most basic, translation is the practice of making texts or speech available to those who do not read or speak the original language. It is connected to the goals of expanding access to ideas, art, or emotions beyond their original context. Translation need not only carry ideas across language; we can also translate across historical contexts by bringing ideas from a foreign country of the past into the familiarity of the present. Therefore, translation is not only linguistic but also cultural, spatial, and historical.

To make these leaps across varied contexts, translation requires recursively moving from the original to the translation and then back to reconsider the original. One outcome of this process is that meanings from the original text may not work in the same manner in the translated language because words do not carry precisely the same meanings across contexts. A translator’s job is to navigate across these gaps in meaning, determining the best path forward when each choice is imperfect. To do so, translators make pragmatic choices. Then, they make them work.

As may be clear, the practice of translation is characterized by contradictory demands. On the one hand, translation is inescapable if people are going to understand one another across language and lifeworld. It also has tangible benefits that go beyond mutual understanding: translation enables conceptual and emotional enrichment as people learn about places and times unlike their own, enabling them to integrate those experiences into their own lives.

On the other hand, translation inherently involves changes in meaning; any translation is necessarily wrong because it is impossible to precisely reproduce the original. The problems are myriad (see Alter 2019 for examples of each of the following challenges). At a basic level, the two languages may not have equivalent words. Deeper in the linguistic structure, the original language may have patterns, structures, or rhythms that cannot be reproduced in the destination language (see Croft 2022 for an excellent discussion). Sometimes the multiplicity of intended meanings might be lost as linguistic play such as puns may not be reproducible across contexts. Beyond the linguistic problems, the text’s original meaning may be intentionally ambiguous, which may be difficult to reproduce if a destination word does not have the same ambiguity (Venuti 2019).

All of this suggests that translation is more than a mechanical exercise. Translation is simultaneously an interpretive, pragmatic, and creative craft bearing the hallmarks of the translator’s voice and sensibilities, perhaps as much as the original author’s (Alter 2022). In this sense, the craft of translation is similar to other crafts: it follows repeated patterns of practice, not because they follow rules set in stone, but because they work. The judgment of the person doing the craft is crucial as they adhere to rules and make choices that help the craft evolve depending on new goals or ambitions. Therefore, the process of translation is subjective and dependent upon an array of assumptions about authorial intent, a translator’s linguistic and cultural capabilities, and the social and historical conditions under which a translation is written. Put differently, translation is a deeply human, rather than a technical, exercise.

Each of these challenges is magnified because there is always a gap in our understanding of an author’s intent, the ability of a destination language to express that intent, and the need to navigate across that divide. It also suggests the need for many translations, as no single translation can ever be “correct.” Even with multiple iterations of a text to express its complex fullness, readers may have good faith disagreements about the text’s meaning.

To take these insights from the practice of professional translators into the study of politics, scholars cannot be removed from the research—they are a piece of it and will have been changed by it. By translating, we can alter our understanding of our own language: when we acknowledge and see the similarities, differences, ambiguities, and nuances through the practice of translation, we challenge our taken-for-granted assumptions about the world. Such shifts, we argue in the remainder of the paper, characterize some of the best social science. Thinking like a translator allows us to see how moving back and forth between research sites deepens our understandings of each, allowing us to see them in new ways and with new lenses, all of which transform the kinds of empirical and theoretical insights we can produce—possibilities to which we turn in the next section.

TRANSLATION IN POLITICAL SCIENCE

If struggling to make an idea work across contexts is a core part of a translator’s practice, it is a challenge that should be familiar to students of politics, as our cases rarely fit our models perfectly. But the practice of translation may also offer a path forward by recasting the process of theorization as a craft where a scholar must make pragmatic, creative choices about how to extend a theory when prevailing rules do not work well. Translation allows us to focus on the places where cases do not map neatly onto one another but where we still see enough that is similar to bring them into conversation and develop theory from resulting frictions. It then allows us to take those theoretical insights to other cases, creating space for them to speak to other times and places without needing perfect overlap. ¹² The embrace of difference and the emphasis on context means we no longer need causal chains to be replicated, or concepts to work in the same way, or even to speak to the same outcome to claim that our insights speak to other cases. If we allow ourselves to see how the same processes and concepts can be refracted through local contexts, we can think about how particular elements of arguments might generate insights even as all components of the original argument might not speak to other cases.

Here, as with literary translation, translation for political science places an emphasis on an arguably underappreciated element of our work: creativity. We noted that literary translation is necessarily creative because translators confront phrases, ideas, or wordplay that do not carry neatly into a destination language—a process that forces creativity when there are few rules. The same is true of our scholarly work, perhaps most obviously (though certainly not only) when we develop theory. As we discuss in our third contribution, elaborated below, in contrast to prevailing models of empirical observation, inductive hypothesizing, and deductive hypothesis testing, theory development can be reconceived as taking place through an abductive process whereby scholars see mismatches between an empirical or conceptual issue and the theoretical priors they carry; then they creatively reshape both their theories, the underlying concepts they deploy, and the underlying research questions themselves to account for the empirical surprises they encounter (Schwartz-Shea and Yanow 2012, 26–33). This suggests that theory development is more than a technical exercise because it requires the scholar to see both the empirical and conceptual objects anew, a process that depends on a scholar’s creativity to reshape a theory for a new context.

Our emphasis on the importance of context in cross-case comparison is not new (e.g. Locke and Thelen 1995; Simmons and Smith 2021). In particular, the focus on analytically parallel, yet empirically different (and therefore seemingly incomparable) phenomena that characterizes contextualized comparison (Locke and Thelen 1995) allows for theory building out of empirical frictions similar to those we highlight with translation. However, this literature rarely speaks to questions of how the findings from a contextualized comparison generalize. If the contextualized comparative approach encourages us to build individual research designs around different yet analytically parallel phenomena, translation applies this same principle to thinking about how our research applies beyond our studies. To borrow from Locke and Thelen, one way of thinking about what we are proposing might be to consider translation as a form of “contextualized generalizability” (see for example how Simmons’s argument has been translated below).

In the remainder of this section, we show how the practice of translation could impact how scholars approach three key stages of the research process: how we build and deploy concepts to order the world, how we conceive of generalizing an argument so that an entire causal sequence does not need to carry to other cases, and what it means to build knowledge by shifting to an abductive logic of inquiry. In sum, the practice of translation pushes us to think about the study of politics as a craft, as opposed to a set of invariable rule-bound techniques—a change that can productively shift how we conceptualize our research practice altogether.

How Translation Changes Conceptualization

The practice of translation can help us rethink how we develop concepts by pushing us to reflect on them from the specific contexts we study, taking a situated, rather than abstract, approach to conceptualization. By taking contextual experiences and translating them back to our received ideas, translation can reveal our taken-for-granted assumptions about the world and potentially change them in the process.

Dominant approaches to conceptualization—particularly Sartori’s (1970) work on the “ladder of abstraction”—privilege the development of abstract, encompassing concepts to capture as many cases as possible without “overstretching.” Sartori (1970, 1035) notes a tension in this approach, though, arguing, “it appears that we can cover more—in traveling terms—only by saying less, and by saying less in a far less precise manner.” This critique is foundational to our own intervention, but we argue that instead of turning toward taxonomies, which Sartori (1970, 1039) argues provide “an orderly series of well-sharpened categories, and thereby the basis for collecting adequately precise information,” we should examine what actors do when they use concepts in practice and remain constantly open to rethinking the concepts to which our cases speak.

With this shift, the translation would seek to develop concepts by refracting them through their use in specific times and places to recursively reflect on prevailing concepts as they are deployed within the discipline, potentially generating new insights about political processes (Becker 1998; Schaffer 2015; see also Kreuzer 2023, 85–90). By situating concepts within local contexts, scholars can engage with similarities and differences across cases (including within the discipline of political science as a “case” in its own right), allowing us to see how elements of a concept both carry and are transformed as they are used in different places. The consequence would be to shift away from whether a concept is “misformed” (Sartori 1970) and instead ask how concepts work (Wedeen 2004). Here, in specific contexts, we can either see how people deploy concepts in their daily lives (potentially in surprising ways) or understand the ways in which their lived experiences of a political process challenge our prevailing conceptual abstractions (or potentially develop new ones). To see how a concept works in practice is not to move up or down a ladder of abstraction; it is to move off the ladder altogether onto the ground.

A good example of how the practice of translation differs from existing approaches to conceptualization is the study of democracy. It might appear that, by refining concepts through situated, contextual analysis, recent efforts to conceptualize varieties of democracy do the translational work that we suggest (Coppedge et al. 2011). These efforts do important work, allowing us to move down Sartori’s ladder of abstraction and see democracy as having multiple instantiations beyond the classic high-level abstraction associated with minimalist conceptions (Przeworski 1999). However, to think about how democracy translates across contexts is to do something different: to look at usage in context and then reformulate that broad concept on the basis of its use in particular cases (Schaffer 2015).

Nicholas Rush Smith’s (2019) study of democracy and vigilantism in South Africa illustrates how one can situate political scientific concepts in local contexts, reflect on them from that situated perspective, and then deepen the prevailing concepts as a result. Smith conducted ethnographic and archival research to understand why South Africans frequently turned to vigilante violence to punish alleged crimes, rather than turning to South Africa’s newly democratized state. During his interviews, he found that a major enabling condition for vigilante violence was ambivalent feelings about the country’s democratic rights dispensation.

Practitioners of vigilantism told him that while they appreciated the country’s constitutional rights in the abstract, when put into practice in their neighborhoods, they saw rights as enabling crime and insecurity. Rights to due process were of particular concern, as vigilante citizens claimed that the rights alleged criminals were legally due might see them released following arrest where, vigilantes claimed, they could continue endangering the rights of other citizens.

Smith then sought to understand how this interpretation of rights pushed them to consider engaging in vigilante violence. He translated these vigilante citizens’ concerns back onto the abstract models of constitutional rights and democracy prevailing in the discipline. In various ranking programs, rights feature as a crucial component of a well-functioning, high-quality democracy. In these terms, an effective democracy would consistently uphold citizens’ rights. But in South Africa, vigilante citizens held the opposite view: when the South African state effectively upheld alleged criminals’ rights, the state was “failing,” as this action would enable criminals to continue preying on residents. The effect, for some, was that instead of a new dawn of equal rights, democracy’s “failure” justified vigilante violence.

In terms of the present paper, Smith translated situated concerns about constitutional rights and democracy back onto abstract concepts prevailing in the discipline and showed how rights and democracy could be interpreted differently in a local context with potentially violent results. The analytical effect of this practice is twofold. First, in moving across time and space, translating concepts and descriptions enriches the process of conceptualization by allowing us to see concepts as multifaceted depending on the ways in which they are deployed in context—a practice that differs from moving up or down on a ladder of abstraction to develop more “accurate” or “well-formed” concepts. Second, the practice of translation allows us to center the often-contradictory ways in which our interlocutors conceptualize the world that can expose potential contradictions in the prevailing concepts we deploy in the discipline with the possibility to deepen, enrich, and even challenge them.

How Translation Changes What It Means to Generalize an Argument

The concept of translation also points to a potentially larger claim about how to think about what a generalizable explanation is in the first place. If qualitative scholars often assume that claiming their argument generalizes necessarily involves moving the entire causal chain to other cases, the translation model suggests that such processes could be broken down into smaller components where only certain parts carry over. These smaller components could be used to help explain outcomes or processes entirely different from those in the original analysis. This points to our second argument about the value of translation: the practice can show how parts of an argument translate to other cases, even if the whole causal chain does not—an approach to making general arguments that differ from existing approaches that usually look to see every link, or combination of links, in a causal chain replicated elsewhere.

This suggests a broader potential insight: if a process of translation assumes that not every part of an argument has to carry across cases for them to speak to one another, the practice of translation may allow for research goals that are different from generalizing causal inference. Rather, in the translation model, scholars assume that there will be fissures, gaps, and breaks in meaning between the original and the translation, which inhibits the kind of replication in causal mechanisms one would seek when generalizing from discrete cases. The goal is not to test our theories (though other scholars using other approaches could do this work) but rather to bring to light the ways in which, even as processes do not replicate perfectly, the theories we develop to understand them offer insights across multiple contexts.

When we recognize the situatedness of our theories, we start with the assumption that our theories are always partial and imperfect approximations of the social processes we describe—including in the cases from which the theory is derived. Instead of trying to fortify our theories by suggesting they only apply to cases within a set of scope conditions, the practice of translation recognizes that those fortifications are already illusory, presumes the vulnerability of our theories, and seeks to open the theoretical gates that scope conditions close. To reiterate our adaptation of Geertz, we cannot know beforehand where the theoretical insights developed in our cases may be impelled to go. ¹³

Erica Simmons’s (2016) comparative analysis of resistance to market reforms in Bolivia and Mexico provides an example. Her argument focuses on explaining social mobilization in response to marketization in Latin America. Key elements of her theory, however, have been deployed by scholars to explain everything from mobilizations in response to Donald Trump’s election to state support for social subsidies in the Middle East. As they are deployed elsewhere, the component parts of her argument are transformed and combined with other mechanisms. Her argument is not replicated, but elements of it are translated in ways that allow scholars to shed light on widely varied political processes in very different social and political contexts.

Simmons’s project began with a question: what explains communities coming together to resist market reforms? Through a comparison of mobilization against water privatization in Cochabamba, Bolivia, and mobilization against rising corn prices in Mexico, Simmons realized that to understand social movement emergence and dynamics, scholars must focus not only on the political contexts and resources available to potential social movements but also on what people perceive to be at stake during marketization. These perceptions influence mobilization processes and the kinds of groups available for mobilization. Grievances, such as rising corn prices, take on meanings and those meanings matter for mobilization. When people understand markets as threatening not only to material well-being but also to widely shared community relationships, understandings, and commitments, those heightened feelings of group belonging can contribute to broad-based mobilization.

Simmons found that this might be particularly true when subsistence goods are at stake, showing how goods that are at the center of daily life might work to produce similar political effects when people perceive them to be threatened by markets. Because of their central role in daily life, subsistence goods can signify not only “imagined communities” (Anderson 1983) but also what Simmons calls “quotidian communities”—communities that are built through routine and face-to-face interactions. Subsistence threats can tap into these imagined and quotidian identifications, heightening solidarity through bringing to the fore common relationships with the threatened resource. Market-driven threats to subsistence goods can be perceived not only as material threats but also as threats to communities as large as a nation or as small as a neighborhood. Broad-based, widespread resistance can result.

We can break this argument down into several insights that might speak to other cases. We see at least four: (1) grievances take on meanings and those meanings matter for mobilization processes and dynamics; (2) perceived threats to community might play a particularly powerful role in generating mobilization; (3) goods at the center of daily life and practice (subsistence goods) might be particularly likely to produce and signify imagined and/or quotidian communities; and (4) when people perceive subsistence goods to be threatened, they might also perceive their community to be threatened and mobilization could result.

If we require the entire causal chain to replicate, the possibilities for generalization are severely limited. When we break down the argument into its component parts, however, we can think about how these parts can speak to other cases, even if the entire argument (or even the components) does not replicate precisely. The range of insights available in the argument can inform how scholars think about a variety of processes, some related to social movements, and some not. By showing how a few scholars have translated two of Simmons’s insights not only to very different kinds of social movements in very different contexts but also to processes entirely unrelated to social mobilization, we aim to demonstrate how this process of translating component parts to other cases works and why it is valuable.

Perhaps the most widely generalizable insight from Simmons’s work is the idea that grievances take on meanings and those meanings matter in social mobilization processes. McKane and McCammon’s (2018) research on the 2017 Women’s Marches offers a particularly good example of how components of Simmons’s argument translate to mobilizations that look very different from those in her original analysis. McKane and McCammon (2018, 402) asked why some locations were home to large marches, while other locations had smaller marches or none at all. Citing Simmons, they found that it is impossible to understand this variation without understanding the variety of meanings that the grievance—Donald Trump’s election to the presidency—took on for people throughout the country. They showed that even as the election outcome was simultaneously imposed, widely felt, and led to strong grievances, the meanings it took on with respect to feminist outrage, racial injustice, poverty, and immigration (among others) help us better understand not only why people participated but also why they did not.

The different meanings the grievance took on created challenging circumstances for movement organizers (McKane and McCammon 2018, 418) who had to navigate mobilization across those meanings. Meanings related to white feminist outrage dominated early organizing but marginalized other meanings that the election took on for other groups. This dynamic, McKane and McCammon argued, helps us better understand why Women’s Marches were less likely to occur in districts with larger Black populations. McKane and McCammon applied Simmons’s insight that the meanings grievances take on matter; in Simmons’s case, the various meanings that water or corn took on helped mobilize broad-based constituencies, while with the Women’s Marches, multiple meanings created challenges for mobilizers. McKane and McCammon translated one element of Simmons’s argument—but not the whole causal sequence—to a different time and place, showing how Simmons’s insights help us better understand not only movement unity but also movement division.

Second, the idea that subsistence goods can take on community-related meanings might apply to any number of political processes where subsistence is at play; the power of the contribution is not limited to processes of social mobilization. In his book on the politics of bread in Jordan, Martínez (2022) translated the theoretical insight to Jordan to show how the meanings that bread takes on are imbricated in understandings of community, citizenship, and state obligation (Martínez 2022, 223). Bread, he argued, means more than its financial value or its value for physical sustenance. Thus, not only the good but also the welfare program that helps distribute it elicits “attachments and anxieties” that differ from those of other social welfare policies. Citing Simmons, Martínez takes this one piece of her argument, which in her context helps explain social mobilization, and instead uses it to better understand everyday moral economies and relationships with welfare policies. He connects the community-related meanings of subsistence to “state performances, the imaginaries they espouse and the conceptions of citizenship they work to entrench” (Martínez 2022, 224). These understandings then help explain why Jordanian policymakers have maintained arguably inefficient welfare support of bread, even as they have cut many other social programs. Effectively, Martínez translates one element of Simmons’s broader argument not only to bread and to Jordan but also to help us better understand state policymaking—a far cry from the social mobilizations around water and corn at the core of Simmons’s empirical work.

Even as every element of Simmons’s causal chain is unlikely to transfer precisely outside of her cases, as the examples here show, many of the insights she develops can be (and have been) translated to a wide range of political phenomena. If scholars looked to whether Simmons’s argument replicated, they would likely miss all the ways in which her insights inform our understandings of processes as disparate (and seemingly unrelated to her original cases) as political marches in the United States and the politics of the welfare state in Jordan.

In looking at the ways Simmons’s argument can be disaggregated and translated to help us understand entirely different political processes, we see how translation creates space for different goals for the research process—goals that situate us deeply in context yet allow us to directly engage with how our arguments speak to other times and places. The concept of translation, then, creates space for us to think differently about how we answer the oft-asked question: “How does your argument generalize?” We can now think about how component elements of an argument translate to other contexts, even if they do not replicate perfectly, and offer crucial insight into political processes outside of the original focus of our research.

How Translation Allows Us to Rethink the Process and Meaning of Knowledge Accumulation

This possibility of breaking down theories into component pieces has implications for a final part of the research process: building knowledge and creating theory. Currently, for many qualitative researchers, knowledge accumulation happens as case studies pile on top of one another, pushing us to either discard existing theories or set new scope conditions for them. Scholars often draw on many studies to abstract out a broad causal process for a given outcome. For example, scholars might develop general theories of why revolutions occur by stacking findings from many different studies to aggregate them into a synthetic account that offers a comprehensive explanation of revolutions (e.g., Goldstone 2003; Mahoney 2003). When scholars find their cases don’t fit the general model, they then narrow their arguments and tighten the scope conditions around their theory, or they claim they have falsified the old theory and toss it out altogether.

We find this model of knowledge accumulation unsatisfying for two reasons. First, despite commitments to context in case description, when cases are aggregated for the purposes of theory development, the unintentional effect is to strip those cases from their contexts and freeze history and geography. Second, as Tavory and Timmermans (2014, 6) argue, this model fails to account for how new theories emerge. Observing that something is empirically interesting, for example, depends on prior theoretical assumptions because scholars cannot know what demands explanation without theoretical priors that are confounded by a given empirical observation. Empirical observation and theorization are intertwined, not separate moments in a research process.

What might an alternative model of knowledge accumulation look like? We argue that the practice of translation, founded in a recursive, abductive logic of theory generation, more closely approximates how scholars work and more effectively accounts for how scholars develop theory than a linear process of knowledge accumulation. Abduction refers to the process of developing a theory to explain a surprising empirical phenomenon when that theory falls short. The primary goal is to develop a new theory to account for unexplained phenomena, which differs from the emphasis placed on empirical testing and falsification in the “normal” scientific process (Tavory and Timmermans 2014, 36). ¹⁴ In this model, scholars often creatively reshape their theories, the underlying concepts they deploy, and the underlying puzzles themselves to account for empirical surprises (Schwartz-Shea and Yanow 2012, 26–33).

Conceiving of this process as one of translation fronts the communal, intersubjective nature of theoretical innovation, as the process of identifying empirical surprises necessarily depends on a scholar’s familiarity with prior theory and sets the goal of innovating on it to explain empirical surprises—a process rooted in a community of scholars reading one another’s work. It also fronts the creative work of theory development, as the craft requires the scholar to see both the empirical and conceptual objects anew, a process that depends on a scholar’s creativity to reshape a theory for a new context.

As an example of what we mean, take the ways in which scholars have engaged Tilly’s (1990) classic argument that war makes states—a set of arguments originally derived from the rise of centralized states in early modern Europe. As scholars have engaged Tilly’s argument to understand other world regions, they have found that changes in the underlying process of state formation have produced major changes in terms of both how war makes states and the types of states that war makes.

Dan Slater’s (2010) work on state formation in postcolonial Southeast Asia is a good example. Explicitly building on Tilly, Slater translated his model to Southeast Asia by taking a component of Tilly’s theory—the broad relationship between war-making and state-making—and creatively adapted it to a new geographic and historic context. However, where Tilly focused on the ways in which external war made European states, Southeast Asia had a surfeit of internal rebellions, coups, and secession attempts. In the differing patterns of war (external versus internal), Slater found that civil war—not international war—made Southeast Asian states and that the type of internal war affected the type of state. The upshot is that because of this lack of equivalence, Slater tells us something new about war-making and state-making: contrary to the view that civil war is inherently destructive, certain types of internal conflict can make states, too.

However, scholars have not only examined the ways in which the process of war-making leads to centralized states. They have also asked what kinds of institutional arrangements war can make. Rachel Schwartz’s (2023) research on war and state-making in Central America is an apt example. Citing both, Schwartz translates Tilly’s and Slater’s models of state formation to a new context: civil wars in Guatemala and Nicaragua. She found a previously unidentified institutional model: “undermining institutions” or durable institutional arrangements that nonetheless distort routine state activities. As do Tilly and Slater, Schwartz suggested that war enables institutional innovation. But instead of developing the bureaucratized routines that are Tilly’s and Slater’s focus, Schwartz found that Central America’s civil wars created patterns of behavior by state functionaries that undermined formal rules, as the individuals charged with counterinsurgency amassed discretionary powers that they continued to deploy to corrupt ends after the civil war was over. The effects of Central America’s counterinsurgencies, Schwartz argued, are durable institutional patterns of elite actors having enormous discretionary power that forms the bedrock of the postwar political order, even as it undermines the state from within.

Both Slater and Schwartz engage in theoretical innovation—identifying a new pattern of war-making and state-making and naming a type of institutional arrangement that did not fit existing expectations. Yet, neither of their contributions neatly fits into the dominant model of knowledge accumulation.

This highlights a key difference between a logic of linear knowledge accumulation and the abductive logic that underpins translation. As noted, the former pushes scholars to create abstract hypotheses and to empirically test them, or failing that, to set scope conditions for causal sequences. But neither Slater nor Schwartz empirically test Tilly’s argument, nor do they falsify it, as each credits Tilly’s claim that war created centralized states in Europe. Slater and Schwartz do not develop a broader abstract model of the relationship between war-making and state-making that could reasonably be called a generalization of Tilly’s findings (at least if generalization is understood as replicating causal mechanisms across cases) because each finds frictions or discontinuities between the Tilly model and the empirical realities in their respective field sites. They do not set scope conditions for Tilly’s argument, given that they would presumably agree that external war can build centralized states but that this pattern is simply not meaningful in their regions of interest. Similarly, they do not suggest that Tilly’s argument (or their own arguments) operates strictly at the middle range. They broaden Tilly’s original claim that war-making leads to state-making in new geographic areas, but they find different patterns in the relationship, which undercuts the ability to generalize the specific external war-making and state-making process Tilly identified. While Tilly, Slater, and Schwartz are all clearly participating in a similar research endeavor, they are not accumulating knowledge through a clear, linear cycle of empirical investigation, inductive hypothesizing, deductive testing, and the falsification of the hypothesis. Something else is happening.

This process can be reconceived as a recursive, abductive process of translation. Each is taking insights developed in one context, applying them to a different context, taking note of the friction across the contexts, and developing new theories and concepts based on this friction—much as when literary translators identify frictions among languages to reveal something subtly new about the meaning of a text in translation. Put differently, they are engaged in a distinctive kind of theoretical practice from that prescribed by the inductive hypothesizing and deductive testing model: Slater and Schwartz extend Tilly’s theory, even as they reinvent it to explain local conditions and make their context-dependent empirical surprises less puzzling. For example, Schwartz develops a new concept to enable this theoretical reinvention. Therefore, translation—the idea that they are abductively extending and recreating Tilly’s argument as part of theory building—captures the process more effectively than the language of replication, falsification, scope conditions, or mid-range theory.

Why is it important to recognize that the practice of translation leads to a different approach to inquiry than the prevailing inductive/deductive model of knowledge accumulation, particularly if scholars are already engaged in this practice? First, as noted, recognizing that much important research takes place through an abductive process where theory development, rather than empirical testing, is a prized outcome is important to allowing emerging scholars to pursue such research in their own work. More radically, though, seeing theoretical development through the lens of translation may push us to rethink the idea of knowledge accumulation altogether.

Contrary to prevailing views of knowledge accumulation in which the system of discovery creates a more unified view of the world, here we have the opposite: a model that centers tensions between broad theories and the specific empirical contexts to which they are applied as the primary sites of theoretical discovery. Tilly, Slater, and Schwartz all agree that war makes states, for example, but none of them finds that war makes states in the same way, nor that war makes the same kinds of states. Yet this mismatch is theoretically generative, allowing Slater and Schwartz to completely rethink the ways in which wars make states while extending Tilly’s core idea in unexpected ways. This suggests that instead of testing to allow theoretical generalization or to determine the scope conditions for an argument, a key goal of this practice is privileging the frictions discovered in the abductive process of theory development to create new ways of seeing the world—much as an effective literary translation can open new conceptual, emotional, and political vistas.

ASSESSING THE QUALITY OF A TRANSLATION

All of this leaves a key question unresolved: how do we evaluate the quality of a translator’s work? We see two broad standards: empirical and creative. The empirical standard is, at first glance, the more straightforward. It essentially relies on whether the translator got it “right.” For professional translators, this would include word choice—did they faithfully choose words, sentences, and phrases that reproduce the meaning of the original text (Smith 2018)? Similarly, for a political scientist, we want to know if scholars provide an “accurate” account of the political processes they study.

How does one know that a scholar got the story right? In truth, this problem is not new, and existing solutions to verifying the accuracy of a scholar’s claims would work equally well for translation. These include classic practices such as checking with experts who know the cases, as during peer review (Schwedler, Simmons, and Smith 2019) or having scholars publish the underlying source material including interview transcripts or historical documents so that scholars can “trust but verify” one another’s claims (Elman and Kapiszewski 2014; 2018).

However, translation differs from existing verification standards in its emphasis on creativity—a value that may come into conflict with the accuracy standard. Professional translators have had robust debates about the degree to which a translation can be “disobedient”—that is, purposefully depart from the original text—and still be a quality translation (Kaza 2017; Smith 2018). Some translators claim that translation is an inherently creative endeavor that requires license because they cannot mechanistically choose words to represent the original language; the interpretive nature of their work guarantees deviation from the original in some sense.

Something similar, we suggest, is true for political science. If we recognize that reproducing, say, an entire causal chain does not work across cases but that our original case has something to say to cases beyond its scope conditions, we might consider the creativity with which a scholar makes the causal argument extend to other cases—echoing Geertz above—to be a useful standard. Having creativity as a standard is not cost-free, nor is it neutral. One major cost is that the drive to creativity may conflict with the drive to accuracy. Anderson’s (1983) work on nationalism—an argument whose creative conceptualization of nations as “imagined communities” has been one of the world’s most powerful social science concepts—is sometimes criticized by area specialists as having empirical flaws (Cheesman 2021, 65; Pepinsky 2022, 1395). In such instances, how to balance the accuracy and the creativity standards would ultimately have to be debated by scholars.

Such debate points to a third way of assessing the quality of research in translation: the debate it produces. If we accept that there might be multiple standards and that they may conflict, then we must also accept that there will be good faith disagreement among scholars who have different values—say, a work’s empirical richness versus its theoretical elegance. Such debates are not new. However, the translation approach would value this debate as primary, recognizing that the debate is the thing itself when it comes to setting standards. Relying on, say, a confidence interval to assess a model’s validity or parsing a segment of an interview transcript to determine whether an argument is accurate is to arguably foreclose debate by reducing assessment to technical parameters. The goal of a successful translation should be to open a debate, foster conversation, and provoke good faith disagreement because it is through the practice of debate that we learn.

While some might argue that a standard based on debate is inherently indeterminate and, therefore, flawed, we would counter that while such a standard is indeed indeterminate, it is superior for it. There are at least three virtues that the primacy of debate and the clash over the relative valuation of facticity and creativity have over a strictly technical standard. The first is that valuing debate lives up to “traditional” scientific goals, at least to the degree that the scientific method is premised on debate over how scientists produce and interpret data. Second, valuing debate is humanistic, to the degree that it values multiple perspectives as being potentially “correct” depending on the vantage point from which one views the standard. Finally, valuing debate is democratic, to the degree that one understands democracy as a process of public deliberation and disagreement. Emphasizing the democratic aspects of these standards is also to recognize the politics inherent in political scientific research.

Adopting the politics of translation forces political scientists to consider the historical and intellectual context in which they are working and the political assumptions under which they are developing generalizable claims. In its ideal form, the logic that underpins generalization is presented as an essentially neutral activity—a technical application of data and theory to understand how some aspect of the political world works. As King, Keohane, and Verba (1994, 9) famously argued about social scientific research, “The content is the method.” However, we would suggest that generalization is a political activity, albeit not one that appears as such. The epistemological assumptions brought to bear on reading data have a history that necessarily shapes what theories are produced, even if the process of theorizing can erase this history by appearing neutral (Riofrancos 2021).

In beginning with the premise that there is a politics to knowledge, which might not be readily apparent translation requires scholars to foreground the politics that underpin their work. In forcing scholars to tack back and forth between their original ideas, the world they engage, and the frames they bring to the world, translation forces scholars to ask several challenging questions: Whose ideas, concepts, and beliefs are we generalizing? From what historical, political, and cultural background do those ideas and theories emerge? To what degree are these concepts taken for granted and, therefore, not adequately reflected on? What are the political and intellectual effects of our not reflecting on this background? And, when we do not reflect on these concepts, what alternative ways of seeing the world might we miss?

Asking such questions delves into thorny intellectual and political territory—something with which professional translators regularly grapple. Indeed, for some, translation is an act of cultural imperialism in which the authenticity of an original text is sullied by appropriation (Apter 2021). Something similar might be said of political science: it takes predominantly Euro-American concepts and transposes them onto the world (Cheesman 2021).

However, another way to understand the politics of translation and political science is premised on the idea of allowing the spirit of human interconnectedness to flourish. If we start with the assumption that we translate because we want to give others access to an idea or experience, we can see translation, in Kaza’s (2017) words, as an “act of hospitality” that “recognizes both the dignity and the difference of the other.” The metaphor of hospitality is particularly apt because, as Kaza points out, it “acknowledges that the host, too, will have to be changed by the encounter. We may have to unravel and remake ourselves with others.” Translation assumes interconnection, openness, and a change to both the original and the new version that may enrich our understanding of the messy human experience by considering each text in light of another. Were we to adopt a similar set of assumptions about the concepts with which most political scientists are trained in Euro-American graduate schools, we might be able to allow similar openness that changes how we see the world.