A tax by any other name? Conceptions of taxation and implications for research

Abstract As taxation has become a prominent issue on the international development policy agenda, a growing body of research has focused on taxpayer perceptions and experiences of taxation. A strand of this research emphasises the importance of the historical, political and social context of taxation. We position ourselves in line with this research as we pay attention to the emic definitions of taxation in Africa across contexts, languages, and time periods. We explore how the conception of taxation in different contexts is closely interrelated with the language used to describe it, with language being a product of histories of colonialism, conflict, and extraction by social, traditional and political actors. We argue that studies of taxation, particularly survey-based research, need to be complemented, if not informed, by a deeper understanding of the diversity of tax landscapes and of the meanings ascribed to taxation in a given context. This will strengthen content and interpretive validity of taxpayer perception data as well as provide important nuances to the understanding of the dynamics of taxpayers’ experiences of contemporary states and systems of taxation.

taxation in a given context. This will strengthen content and interpretive validity of taxpayer perception data as well as provide important nuances to the understanding of the dynamics of taxpayers' experiences of contemporary states and systems of taxation.

I N T R O D U C T I O N
As taxation has become a prominent issue on the international development policy agenda, a growing body of research has focused on taxpayers' perceptions and experiences of taxation using cross-country comparative survey data (Levi &  This is in line with the broader quantitative turn in both political science and African studies more generally (Cheeseman ; Cheeseman et al. ) and reflects in part the greater availability of tax-related data through crosscountry surveys such as the World Value Survey and the Afrobarometer. Undeniably, much has been gained through the increasing accessibility of cross-country survey data. We argue, however, that insufficient attention has been paid to issues related to content validity and the cross-context comparability of key tax-related concepts that are at the heart of analyses of a wide range of topics, including relationships between the willingness to pay taxes and perceptions of corruption, state legitimacy and public service provision. While 'serious questions have been raised about the quality of the data used in some quantitative work in African studies (Cheeseman et al. : ; see also Jerven , ), little attention has been given to the specific problems related to studying taxation through cross-country survey data. This is despite both proliferating research in this area and preliminary evidence of validity issues. Recent research on the comparability of different measures of tax morale, for instance, makes clear that how questions about tax are asked influences the answers (Prichard ).
With this briefing, we aim to encourage critical reflection concerning the use of tax-related data across diverse contexts.  We explore the ways in which diverse experiences and languages may influence the meaning of taxation. Essentially, we show that a tax by any other name does not retain the same meaning across contexts. We do so both by drawing attention to the ways in which history and taxpayer experiences shape the meaning of taxation in any given context and by comparing how the meaning of tax-related questions changes across different languages in large-scale cross-country surveys, such as the Afrobarometer. We thus position ourselves in line with a growing body of research that emphasises the importance of the historical, political and social context of taxation as we explore the emic definitions of tax across contexts and languages (Prichard & van den Boogaard ; Meagher ; Bak ).
We show that the concept of taxation takes diverse forms across different contexts as it is shaped by the experiences of taxpayers and the language used to describe it, with language being a product of histories of colonialism, conflict and extraction by social, traditional and political actors. The reality of public finance is often much different than that assumed by researchers, who tend to approach taxation with a bias towards the state and the national government in particular. By contrast, many individuals in African countries have limited experience of direct taxation by central governments, engaging more with local government taxes and user fees, informal taxes and non-state tax collectors, including traditional authorities. The diversity of tax systems across African countries is stark and shaped to a certain extent by variation in colonial extractive practices. Moreover, the diversity of languages within and across countries in Africa shapes the meaning ascribed to key concepts such as 'taxation', 'tax authorities' and 'tax morale' across sub-national regions and countries, in line with the idea that 'the meanings of words consist in the actual ways in which they are used in various contexts' (Schaffer : ; see also Wittgenstein ). Without recognising this diversity and interrogating the meaning of concepts in a given context, research on taxation in Africa risks conflating diverse ideas and experiences, misrepresenting taxpayer perceptions, and drawing ungrounded conclusions from comparative analyses.
We argue that studies of taxation, particularly survey-based research, need to be complemented, if not informed, by a deeper understanding of the meanings ascribed to taxation in a given context. This may be achieved through a greater prioritisation of conceptual analysis, mixed methods approaches and local language research. This will help to strengthen measurement and interpretive validity of taxpayer perception data, as well as provide nuance to the understanding of variations in and dynamics of taxpayers' experience and interactions with tax systems.
Everyday experiences of taxation are often very different than assumed by analysts studying a country's 'tax system'. There are many ways in which the 'tax system' may be lost in translation and in which researchers' understanding thereof could be misaligned with that of taxpayers. Failing to pay attention to how these experiences and meanings differ has implications for how we use and interpret survey data and, ultimately, the conclusions we draw. We point to four ways through which everyday realities may not be captured by broad questions about 'taxation'.
First, in low-income African countries, few people have more than a minimal experience with the 'tax system' referenced in surveys and commonly interpreted to mean the national tax authority or administration, in line with a common bias towards national statistics when it comes to taxation (Moore et al. : ). Instead, taxpayers' primary encounters are with local levels of the state and tax systems. Evidence shows that direct taxes make up a small proportion of tax mobilisation as a whole (Moore et al. ; ICTD/UNU-WIDER ; van den Boogaard ), while taxpayer surveys across a range of contexts show that a large proportion of the population do not pay direct taxes to the government, either because their incomes are too low or they are otherwise able to avoid paying.  The greater prevalence of local taxesand the proliferation of small and 'nuisance' taxes (Fjeldstad &  . Given this, it seems likely that citizens may consider a broad scope of payments as part of the tax landscape making up the 'tax system'. Though some analysts emphasise that taxes and user fees should not be conflated, for taxpayers 'the distinction is far less meaningful as taxes and user fees often appear as relatively interchangeable costs of financing local public goods' (van den Boogaard : ).
Third, much research and theoretical thinking about public finance tends to exhibit a 'statist' bias, while the reality in many low-income African countries is that people contribute to the financing of public goods to both state and nonstate actors through informal taxationthat is, non-market payments that are not defined or enforced by state law (Guyer  . The practice of taxation and how people experience it is often rooted in informal socio-political institutions and structures, some of which have pre-colonial roots (van den Boogaard & Santoro ). By contrast, the meanings ascribed by researchers to key concepts such as tax and tax authorities are rooted in Western notions of what fiscal systems look like and can fail to comprehend reality on the ground. For example, despite it long being recognised that the meaning of corruption is highly varied and influenced by social, cultural and political institutions and the logics and practical norms of governance (e.g. Olivier de Sardan ; Blundo & Olivier de Sardan ; Nicaise , ), external observers can be quick to assume that informal payments are acts of corruption and will be seen by citizens as illegal. In practice, however, citizens do not always see them as such.
Research suggests that citizens sometimes see informal taxes as fairer than formal taxes (Titeca & Kimanuka ; van den Boogaard et al. ; van den Boogaard & Santoro ), while 'in contexts of weak institutional capacity … the distinction between formal taxes, formal user fees, informal user fees, and illegal payments to state actors is far less clear' (van den Boogaard : ). Accordingly, it is plausible that when asked about the 'tax system', citizens consider the full range of payments, formal and informal, they make to financing public goods. As both the government and the national revenue authorities may be physically and psychologically distant, everyday experiences of taxation are often much more diverse than assumed by analysts studying a country's 'tax system'.
Finally, and following from the predominance of local taxes, user fees and informal taxes, in many African contexts there is a multiplicity of actors involved in taxation. Besides the national or local government, colonial administrative legacies mean that in many contexts chiefs are involved in tax collection on behalf of the government (Balán et al. ; Bolt & Gardner ; van den Boogaard ; Weigel & Ngindu ). When surveys such as Afrobarometer ask for agreement with statements such as, 'The tax authorities always have the right to make people pay tax', how do citizens interpret what is meant by 'tax authorities'?  The fact that 'tax authorities' are plural implies reference to the multiplicity of actors involved, but it is not clear whether researchers take this into account in their analyses. In Sierra Leone, the Krio version of this survey question translates 'tax authorities' as 'those that collect tax' which, in the Sierra Leonean context, is more likely to be chiefs than it is to be the National Revenue Authority (van den Boogaard et al. ). The Diola version from Senegal translates 'tax authorities' to 'mansa', which could be understood simply as referring to 'the state' in the broadest sense of the word, leaving ample room for respondents to interpret the question in different ways and to think of a multiplicity of actors.  Even if researchers conceive of the 'tax system' or 'tax authorities' as referring to the full range of tax institutions involvednational and local, formal and informalthere is likely to be variation in the types of actors that citizens conceive of across countries and sub-national regions. The complexity is exacerbated when recognising that the ways in which taxpayers engage with the tax system is deeply shaped by intersections of gender, class and histories of structural marginalisation (Roitman ; Meagher ; Joshi et al. ). 

L A N G U A G E A N D T H E M E A N I N G O F T A X A T I O N
In addition to the everyday experiences of taxation, evidence suggests that the language used to describe a tax can shape how taxpayers perceive it, as well as their willingness to pay.  Given the great diversity of languages within and across African countries, it is particularly important to pay attention to the terms used to refer to taxation and to recognise that 'different language groups … may have different political vocabularies and conceptions' (Schaffer : ). Translating tax terms into local languages comes with several challenges that may affect their conceptual equivalency.
First, while there are certainly reported cases of "gross mistranslations" (Appiah-Thompson ), many local African languages simply do not have conceptual equivalents for key terms concerning formal taxation and may thus borrow words from a dominant (usually colonial) language. These 'loan' words are then imbued with non-indigenous meaning and may be transformed by their interaction with local cultures, while descriptive language may be used to try to illustrate the meaning. For example, in a question of the Afrobarometer survey referring to 'income tax', the English translation provides a prompt explaining that income tax is 'a tax deducted from your wages by your employer'. While this seems to refer to withholding income tax only, the Krio version of the question asks, 'Do you have work where they pay you? Do they take tax from the money when they pay you?'.  In the Diola version used in Senegal, the question asks, 'If you have a salaried job, would you be willing to pay taxes?'  Another important point is that in Diola, the phrase 'mentère pan ou mangh' translates into 'are you willing to' rather than 'are you required to', which implies a considerably different meaning.
Second, different languages may use different words to describe the same concept. For example, in French, a tax can be described as a 'taxe', 'redevance' or 'impôt', with the former two usually designated for payments that may in English be more akin to a user fee (in that they are typically expected to be requited) and the latter describing a payment made to a general fund and not linked to any particular service or licence. This is further complicated by the reality that there is often considerable variation within countries and across regions in what payments are designated as a 'taxe', 'redevance' or an 'impôt'. In the Democratic Republic of the Congo, for example, these designations are sometimes influenced by the rurality of the region and the authority in charge of tax collection (e.g. the Decentralised Territorial Entities or customary authorities); in some cases, a single word may be used to refer to both 'taxes' and 'impôts', depending on how the terms are understood and accepted in different areas.  In Senegal, survey questions asking about tax sometimes use two words: 'juuti', referring mostly to user and licence fees (redevances), and 'limpo' which is a collective word for all kinds of payments including customs, formal taxes and fees.  In everyday usage, 'juuti' is more often spelled 'duty', revealing the word's relation to English, while 'limpo' comes from the French word for formal taxes 'l'impôt', but has taken on a broader meaning in Wolof. Hence, even among Wolof-speaking respondents in Senegal, the question is likely to be understood in multiple ways.
Third, and in line with the above, language is shaped by context. One clear example is how associations with colonialism affect citizens' perceptions of taxation. For example, in Somalia, where colonial taxation by both the Italians and British was largely based on an extractive (as opposed to contractual) model, formal taxation is today referred to as 'canshuur', meaning 'restriction' or 'burden' and 'signifying a system imposed and regulated by force' (Abshir et al. : ). In Sierra Leone, property tax in some areas of the country today remains closely associated with the colonial hut tax, which was the impetus for a notorious revolt in .  Given this, it is unsurprising that taxpayers and chiefsparticularly in the region where Bai Bureh led the colonial tax resistanceexplicitly, and often proudly, justify their resistance to contemporary property tax by referring to it as akin to the hut tax (van den Boogaard ). Roitman () provides another example of the infusion of colonial experience into contemporary language: in Northern Cameroon, tax (impôt) is spoken of as a price (prix), while the verb taxer in French is used to signify vendors fixing a price on a product at a market. This is rooted in the intimate link between the imposition of the French currency and the fixing of monetary prices on products, which were a means of institutionalising taxation during colonialisation (Roitman : ). Fixing prices ensured creating profits for taxation. Not only was this a central mode of extraction during colonial times, but it carried through into post-independence revenue extraction, including after suspension of formal price controls (Roitman : ). Consequently, the conflation of pricing and taxing lives on in the language and possibly also in conceptions of taxation, even if these conceptions reflect a mix of indigenous and colonial concepts and contemporary politics.
Ultimately, even though the Afrobarometer acknowledges that translation to local languages is 'one of the most critical steps in the entire survey process for obtaining good quality data from our interviews' (Afrobarometer : ), the issues highlighted here may stymie even the best efforts by survey translators. The challenge is further exacerbated by the reality that national surveys such as the Afrobarometer are not available in all local languages. Some local language surveys are certainly better than none, but when some citizens are not able to take the survey at all, there is a risk of sample biases; alternatively, if respondents cannot take the survey in the language with which they are most comfortable, questions or concepts may be misinterpreted. 

I M P L I C A T I O N S
Given the diversity of experiences and language around taxation, we cannot assume that a tax by any other name retains the same meaning. Researchers have to be attentive to how everyday experiences, language and context shape responses to tax-related questions. The uncertainty concerning the conceptual equivalence of survey questions across diverse contexts has several implications for data analysis and interpretation.
For one, the diversity of taxpayers' experiences and perceptions of tax systems will affect content validity, while researcher biasesconcerning most prominently what is a 'tax' and what is a 'tax system'infuse their interpretation of terms, concepts and survey questions (Nolte et al. ). This is a risk particularly, but not exclusively, when researchers are not themselves involved in the design of survey instruments and data collection.  Critically, these biases tend to reflect state-centrism and Weberian institutional ideals. Researchers may believe the idea of taxation to be universal, but the reality of informal and pre-colonial institutions of public finance seriously challenges such assumptions derived from cultural premises and biases that infuse Northern conceptions thereof (Johansson ).

A T A X B Y A N Y O T H E R N A M E ?
At the same time, the flattening of concepts such as 'tax' or 'tax authority' risks leading to a vast oversimplification of the reality on the ground (Cheeseman et al. : ). When survey questions ask about the accountability or transparency of the tax system but fail to probe what is meant by the 'tax system' to citizens, we will not understand how individuals actually interact with systems of public finance and what implications this diversity of interactions has for their perceptions of the state and state-society relations. The problem is thus not just semantic; a poorly grounded understanding of the everyday experiences of taxation in research may inform policies that, at best, have little or no effect and, at worst, erode citizens' willingness to pay taxes and interact with state actors more generally.
Overall, further consideration is needed with regard to how experiences and language affect the validity of research findings. Content validity may be challenged given the contextualised meanings of taxationa problem that is exacerbated in cross-country analyses. Further, the diverse experiences with taxation, historic and contemporary, raise questions about the internal validity of research on determinants of tax compliance as well as on consequences of taxpaying for outcomes such as political participation and perceptions of government performance. For example, survey respondents might respond positively to both the question whether they paid taxes and whether they contacted a public official within the last year; however, as they navigate disparate tax landscapes, interact with various kinds of actors, and have diverse conceptions of taxation, they might do either for widely different reasons. We thus cannot assume that the cause and outcome are linked (here, taxpaying and political participation) or, if they are, are linked by the same mechanism across contexts. For the same reasons, external validity is not given, and the potential for generalising findings from single-country studies to a region or the continent as a whole should be based on positive arguments rather than methodological assumptions. 

W A Y S F O R W A R D F O R T A X R E S E A R C H
As we have shown above, diverse everyday experiences and vernacular related to taxation have implications for interpreting and comparing data across contexts. We argue that studies of taxation, particularly survey-based research, need to be informed and complemented by a deeper understanding of the meanings ascribed to taxation in a given context. In conclusion, we highlight two ways in which researchers can begin to address the challenges outlined above and, in turn, improve the quality of future tax research: first, conceptual analysis and mixed methods approaches and, second, a prioritisation of local knowledge.
A first step towards improving tax research is to take the time to explore the variety of meanings of tax-related terms and concepts in the research context. Conceptual analysis includes a range of methods, including studying translation issues, the structure of concepts, how specific concepts fit into a semantic field of related concepts, evolution of conceptual meanings over time and, not least, how concepts are used in everyday, ordinary language (Schaffer , a, b). A more nuanced understanding of the meanings ascribed to taxation, the tax landscape and taxpayer experiences will inform better survey questions. Importantly, conceptual analysis enables us to begin not 'with an abstract normative standard, but the values of the real actors' (Scott : ). Delineating the real dynamics of taxationincluding the actors and levies involvedfacilitates further precision and disaggregation in survey questions. Conceptual analysis can also provide a better foundation for deciding, for example, whether to use national or local language words for taxation, 'experience-near' language (Geertz ), or colonial 'loan' words. This type of conceptual analysis goes hand-in-hand with being a 'good fieldworker' (Cheeseman et al. : ).
A second pathway to improve tax research is to acknowledge the importance and value of centring locally based research and knowledge. Global inequalities permeate knowledge production within African studies in multiple ways, rooted in history and upheld today, not least by variation in publishing and funding opportunities, including through the funding structures underpinning North-South research collaboration (Obeng-Odoom ; Green ; Mudinga ; Muzalia ; Nyenyezi et al. ). At the same time, the marginalisation of local knowledge is to some extent embedded in different methodological approaches.  There are limits, of course, to what outsiders can understand (Ba ). Even if fluent in a local language, a researcher may not understand the etymology and historicity of concepts, which may be deeply embedded in personal or communal experiences passed through memories, stories, jokes and folklore. We argue, therefore, that it is necessary to forefront local knowledge and include local researchers as co-producers of surveys and analyses. Centring taxpayers' voices and understandings of concepts in research can also be instrumental in addressing conceptual issues (van den Boogaard : ). Ethical considerations here are central, though more genuine prioritisation of local knowledge will serve to improve the design of research and survey questions and the interpretation of results.
Even with the above measures, challenges of content validity may not be entirely overcome. Designing research and surveys involves making choices and weighing trade-offs. We therefore emphasise the need for greater transparency in the design, testing and interpretation of survey-based research. There is often little acknowledgement of underlying issues with data sources (Nolte et al. : ), with 'data users and consumers … often given the impression of a much neater and smoother information surface' regardless of how 'messy it is in the workroom' (Jerven : ). More open discussion of trade-offs, methodological limitations and their implications for the validity of research findings signals thoughtfulness and the credibility of research. Concrete suggestions could involve, for example, including a mapping of actors involved in taxation in a country in an appendix and discussing in-text how survey questions take into account the multiplicity of actors, or including an overview of the names of taxation within or across researched countries and detailing the nuances that may be lost from the choice of survey wording. More generally, researchers should aim to make positive arguments for why we would expect, for example, determinants of tax compliance to generalise beyond the empirical scope of the data, from one context of taxation to another. As we have argued in this briefing, there are many reasons why we should pay more attention to what is in a name and to the variations in how taxpayers experience and interact with tax systems across Africa.

N O T E S
. While these issues are in line with broader challenges related to survey data and content validity, we draw attention to the issues as they relate to taxation, given the growing interest in the field. Likewise, the issues are not distinct to research in African contexts, though they have been less well explored in this context. One recent exception is Struwig & Roberts ().
. For example, van den Boogaard et al. () show that only % of people in northern and eastern Sierra Leone pay any direct tax to the central government; Waris (: ) shows that, with a few exceptions, registered taxpayers make up less than % of registered voters (used in lieu of reliable population data); and Beach () estimates that only .% of Benin's population are registered as taxpayers. More people pay indirect taxes, particularly in the form of valued added tax (VAT), though evidence suggests both that the reach of VAT is more limited than commonly thought (e.g. Bachas et al. ) and that there is considerable variation in the saliency of indirect taxes for taxpayers across countries, depending on the extent to which they are applied to staples and other goods consumed regularly by even the lowest-income groups (Prichard ).
. It is not clear, however, what implications they draw from this knowledge. In their analysis, they continue to use the measure of whether respondents agree with the statement 'The tax department always has the right to make people pay taxes' as an indicator of legitimating beliefs (i.e. quasi-voluntary compliance), and they focus their eventual discussion of the study's limitations on internal and external validity rather than content validity.
. Wording is from Question  from Round  of the Afrobarometer survey in Sierra Leone (). . The statement reads, 'Di wan dem wae dae collect tax get right all tem for mek pipul pay tax'. . The full question in Diola is: 'Sinséne mane mansa koudioume mambi ou thiamale impots.' . It has long been understood that, 'language use, meaning, and standards can also vary with the speaker's class, gender, age, or race' (Schaffer ).
. For example, studies have shown that popular support for wealth and progressive taxes is more likely when they are framed as a 'graduated income tax' rather than as mechanisms to 'tax the rich' (Essig ). In line with this, we see wealth taxes in the context of the Covid- pandemic being widely framed as 'solidarity taxes' rather than 'wealth taxes' per se (Waris ). There are likewise parallels to the language used to describe armed group taxation, with differences in taxpayers' willingness to pay levies described locally as 'protection payments', 'business taxes' or 'revolutionary taxes' (Bandula-Irwin et al. ).
. Notably, the term 'hut tax' is itself illustrative of underlying social and political dynamics; the tax was originally termed the 'house tax', with local taxpayers referring to it as the 'hut tax' to communicate their sense of the unjustness of the levy.
. Even where surveys are available in local languages, the survey may be more likely to be conducted in the language with which the enumerator is most comfortable. Also, identifying the language in which a respondent is most comfortable to speak may be challenging, judging by the different iterations of the question in the Afrobarometer, varying from 'What is your maternal language?', 'What is your language of origin?', 'What is your first language?', and 'What is the primary language you speak in your home now?' These questions are different enough that the result may be respondents taking a survey in a language in which they are not fully fluent.